第三章网络存储技术 Networked Storage

第三章网络存储技术 Networked Storage
网络存储与系统虚拟化技术 Network Storage and System Virtualization Technology 北京大学信息科学技术学院网络与信息系统研究所系统虚拟化及空间信息技术实验室（

本章目标及内容面对不断增长的信息需求和不断发展的网络技术，网络存储应运产生。本章分别介绍了网络存储环境下的直连存储、网络附加存储、光纤存储局域网、IP存储局域网和内容寻址存储，并深入介绍了这些网络存储结构的概念、结构特点、基本原理以及应用环境等方面的知识。本章内容包括5个方面： 3.1 直连存储 (Direct Attached Storage，DAS) 3.2 网络附加存储 (Network Attached Storage，NAS) 3.3 光纤存储局域网络 (Fibre Channel Storage Area Networks, SAN) 3.4 IP存储局域网（IP Storage Area Networks，IP SAN） 3.5 基于对象的存储（Object-based and Storage, OBD） In the previous section, we looked at the components of a storage system. In this section, the emphasis will be on different storage system models-- starting with a basic storage model (DAS) and focusing on networked storage models such as SAN, NAS, and CAS. When effectively designed and implemented, Networked Storage enables more effective utilization of available storage, increased flexibility, and better availability of data. Direct Attached Storage (DAS) - storage devices directly connected host either via an internal or external connection. Network Attached Storage (NAS) - storage devices accessed over a network, in front of the server. Is made up of computing/storage devices running an operating system that has been optimized for file service. Storage Area Network (SAN) - a dedicated network that carries data between computer systems and storage devices. Content Addressed Storage (CAS) – an object-oriented, location independent approach to data storage. It can be implemented using a number of different technologies.

本章目标及内容面对不断增长的信息需求和不断发展的网络技术，网络存储应运产生。本章分别介绍了网络存储环境下的直连存储、网络附加存储、光纤存储局域网、IP存储局域网和内容寻址存储，并深入介绍了这些网络存储结构的概念、结构特点、基本原理以及应用环境等方面的知识。本章内容包括5个方面： 3.1 直连存储 (Direct Attached Storage，DAS) 3.2 网络附加存储 (Network Attached Storage，NAS) 3.3 光纤存储局域网络 (Fibre Channel Storage Area Networks, SAN) 3.4 IP存储局域网（IP Storage Area Networks，IP SAN） 3.5 基于对象的存储（Object-based and Storage, OBD）

Intelligent storage systems
Intelligent storage system models: DAS/NAS/SAN/OBD Networked Storage: NAS/SAN/OBD Intelligent Storage System Front End Back End Physical Disks Cache Host Connectivity Cache

3.1 直连存储(DAS) Direct Attached Storage Internal/External DAS：直接与一台主机相连
Direct Attached Storage, or DAS, is the most basic level of storage. DAS devices either reside as an integrated part of the host computer (e.g., hard drives, removable storage devices, etc.) or directly connected to a single server externally (such as RAID arrays or removable media). As the first widely popular storage model, DAS products still comprise a large share of the installed base of storage systems in today's IT infrastructures. In this module, we will look at what DAS is, its components, management issues, and the challenges that DAS presents.

Internal/External DAS
For the purposes of this module DAS is defined as directly attached storage without the introduction of FC network connected systems, which will be discussed in a later module. One of the features of DAS is that the storage resources are dedicated only to the hosts that are using them. This allows for bottlenecks to storage to be alleviated by adding resources only to pathways that require them. The storage within the array remains partitioned between the systems using the storage (i.e., the data on the devices assigned to the servers/clients is dedicated to the use of that specific directly attached server/client only). Internal External Direct Connect 专用的存储资源：只与一台主机相连。

DAS的优点适合本地数据存储小型存储环境：可以快速部署配置简单，便于部署可靠性高便宜易于管理
Internal DAS is a viable option for small environments because it is relatively easy to deploy and inexpensive in simple configurations. External DAS is ideal for localized data connectivity in environments with a single host or a few hosts. For example, small businesses or departments and workgroups that do not need to share information over long distances or across an enterprise. Small companies traditionally utilize DAS for data serving and , while larger enterprises may leverage DAS for mission critical application data in a data center environment. DAS also offers ease of management and administration since it is either handled by: The server/client’s OS (for internal DAS) A management interface to the intelligent array housing the storage (for external DAS).

DAS的组成 Motherboard Clustered group of processors Processor cards
Connectivity Storage CPU Motherboard Clustered group of processors Processor cards Complete system Internal External Hard disk(s) CD-ROM drive Optical drive Removable media Tape devices/tape library RAID/intelligent array(s) Portable media drives The physical elements of DAS include: CPU Connectivity Storage devices There are many options for each of these elements as shown.

DAS的连接块级别（Block-Level）的访问协议 IDE/ATA 和 SATA SCSI Buss and Tag
主要用于内部连接（Internal）总线 SCSI 并行SCSI（Parallel SCSI）：for Internal 串行SCSI（Serial SCSI）：for External Buss and Tag for External 光纤通道（IBM的ESCON和FICON）的先驱 Logical attributes of Direct Attached Storage (DAS) include the use of block-level access protocols such as ATA (IDE) and SCSI. (1)ESCON——Enterprise Systems Connection，管理系统连接。 IBMS/390的光纤通道，能在60千米距离内以17MB/s速率进行数据传输。ESCON允许外围设备跨大园区和城域分布。与铜基并行总线和特征通道相比，ESCON提供更高的速度并使用串行接口。ESCON定向器是中心和外围连接装置，提供8~16个端口（模式1）或28～60个端口（模式2）。 (2)FICON——FIber Connector，光纤连接器。 1998年和G5服务器一起推出的IBM大型主机通道。它以光纤通道标准为基础，将ESCON的半双工17MB/s传输率提高到了全双工100MB/s。每条FICON通道最高可以支持每秒4000次I/O操作，相当于8条ESCON通道。

DAS的连接：Internal 机箱空间距离连接设备数功耗
In internal DAS architecture, the storage devices are internally connected to a host computer via a physical bus. The physical bus has distance limitations. These limitations are overcome by using a parallel protocol to give faster access to devices. Note: This also results in less voltage utilization (less heat). Most internal buses have a limited number of devices which they can support. 机箱空间距离连接设备数功耗

Parallel Connectivity Cables Serial Connectivity Cable
DAS的连接：Internal设备示例 Parallel Connectivity Cables Serial Connectivity Cable The slide illustrates some common internal bus connectivity cables for devices. Notice that a serial connectivity bus requires fewer strands than a parallel cable due to the serial nature of the protocol. This reduces the number of data pathways, which generally means that less data per second can be transferred. 50-wire SCSI-2 cable 80-wire IDE cable 34-wire floppy cable Serial ATA cable

SCSI, FireWire, Serial ATA, USB, etc
DAS的连接：External Example of an external connectivity cable External DAS HBA HBA SCSI, FireWire, Serial ATA, USB, etc External DAS connects the client device to the external storage device directly via Fibre Channel or other hard connection. External DAS overcomes the following limitations: Internal space of host Number of devices that can be connected Some distance limitations External DAS also: Allows for centralized management of storage devices Facilitates repair and maintenance Easier to swap components due to facilities offered by the array. Note: The Host Bus Adapter (HBA) could be one of many adapters supporting many differing protocols, (e.g. SCSI, FireWire, Serial ATA, USB, etc.). 空间距离设备数管理 Host Storage Device SCSI、FC

DAS的管理：Internal 主要由主机及其 OS进行管理包括：磁盘分区 (卷管理) 文件系统数据寻址
Internal DAS is generally managed through the host and OS or by some third party software. Device management provides many features including: Disk/volume partitioning/management File system specific layouts for the OS Data addressing for storage and retrieval of data

DAS的管理：External 基于阵列的管理可获得性 – multi-path I/O
更低的TCO (Total Cost of Ownership )：管理数据或管理存储设施 A key feature of external DAS management is that the host OS is not directly responsible for any fundamental management of the resources (e.g. LUN creation, filesystem layout, and data addressing). External DAS also introduces the option of multiple pathing to storage resources, although some high-end hosts do offer this internally as well. Internal multi-pathing is more susceptible to system failure. External multi-pathing may be more resilient, depending upon the specific host failure. An additional feature of multi-pathing to storage is the ability to load balance I/O to improve data transfer. A primary cost component for businesses today is managing a multi-vendor storage infrastructure and the data on that infrastructure. Multi-vendor internal DAS storage must be managed individually and by placing the majority of storage externally on single vendor arrays, management becomes more centralized and skill sets required for multi-vendor management can be reduced.

DAS的性能影响 DAS 性能的因素包括： Hard disks Memory cache Virtual memory (paging)
Storage controllers Cache size Protocol supported (e.g. SCSI, FireWire, USB, etc.) RAID level Bus The following may impact DAS performance: Hard disk – Seek time and rotational latency impact performance. The block layout of the stored data also impacts performance with respect to fragmentation. Virtual memory – Paging to disk can slow system due to resource contention (i.e., the virtual memory process competes with applications for disk resources). Storage controller - Controller cache can improve performance. However increasing the number of cache locations can lead to data loss in the event of a disaster if failsafe mechanisms are not implemented at all levels. The type of protocol/interconnectivity must also be taken into consideration for the application. RAID level – Based on the application workload profile, i.e. percentage of read I/O and write I/O , the RAID level can adversely impact performance. For example, in RAID 5 there is a write penalty, so it may not best for applications that require frequent sequential writes. Bus - Higher throughput speed provides better performance, but the distance may exclude specific buses from consideration.

Holds an embedded HBA with an IDE bus connector
Internal DAS示例 Hard Drive 40 Pin Ribbon Cable The motherboard here has an embedded HBA with an IDE bus connector for the cable connection. The cable is attached to this connector. The hard disk is then attached to the other end of the cable. Motherboard Holds an embedded HBA with an IDE bus connector

Cable for external DAS connectivity
Here is an example of the elements that can be used in an external DAS implementation. ESCON HBA Cable for external DAS connectivity

DAS面临的挑战必须与主机直接相连数据可获得性（Availability）的局限数据访问速度降低
较多的单点失效点 bus, multiple path software, host, application, etc 数据访问速度降低 CPU拥塞, caching, multi-pathing 可扩展性（Scalability）的局限与主机的连接端口数可编址的磁盘数距离维护时的停机时间（Downtime） DAS poses several challenges: Hosts must be directly connected. Data availability: Many single points of failure (i.e., bus, multiple path software, host, application, etc.). There is no redundancy or fault tolerance for the existing system. Inability to share data or unused resources with other hosts simultaneously. Scaling is simply a matter of adding more LUNs to the servers attached to the array. The host’s hardware limitations restricts the amount of growth that can be accommodated. For example, the number of ports available on an external array and number of hosts that can be actually connected to an internal bus is limited. Both internal and external DAS have a finite bandwidth available for data I/O to the attached servers. When capacities are being reached, data availability may become compromised. This will have a ripple effect on the performance of all the hosts attached to that specific device or array. Distance limitations of the medium used for connectivity will determine feasibility. Due to the fact that devices are directly attached to the systems, there are few if any server implementations that will allow for drastic device reconfiguration, addition or removal of systems drives, new HBA insertion or removal, etc. without having to first power down the host. This makes scheduled downtime planning and storage provisioning necessary.

小结 Internal/External DAS DAS组成：CPU, Connectivity, Storage devices
DAS连接：块级别（Block-level）的访问协议

本章目标及内容面对不断增长的信息需求和不断发展的网络技术，网络存储应运产生。本章分别介绍了网络存储环境下的直连存储、网络附加存储、光纤存储局域网、IP存储局域网和内容寻址存储，并深入介绍了这些网络存储结构的概念、结构特点、基本原理以及应用环境等方面的知识。本章内容包括5个方面： 3.1 直连存储 (Direct Attached Storage，DAS) 3.2 网络附加存储 (Network Attached Storage，NAS) 3.3 光纤存储局域网络 (Fibre Channel Storage Area Networks, SAN) 3.4 IP存储局域网（IP Storage Area Networks，IP SAN） 3.5 基于对象的存储（Object-based and Storage, OBD）

NAS产生的背景文件共享的需求网络传输速度不断提高 NAS appears Multi-file Copies
Share Directories File Servers: Performance and Scalability bottleneck General OS Commodity hardware 网络传输速度不断提高 NAS appears 硬件：High-performance, scalable hardware 软件：Specialized OS and protocol interfaces designed specifically for file serving. At one time, it was common for users within a given office or department to share directories from their own workstation. As this model continued to grow, data security and integrity became difficult to manage. File servers began to be used as a central repository for users’ data. Over time, as the storage and processing requirements of the file server continued to increase, the operational components of the file server itself became a performance and scalability bottleneck. PC operating systems added a significant amount of overhead, and commodity hardware could no longer meet the requirements for availability, scalability, reliability, and performance of a growing enterprise. NAS devices were developed to address these challenges. NAS devices typically include: High-performance, scalable hardware Specialized operating systems and protocol interfaces designed specifically for file serving. The first generations of NAS devices were typically very high-end and expensive. NAS technology has now moved into the small-medium business market. NAS has also evolved into being more than just a shared storage pool. By integrating intelligent storage features, simplified management, and data protection capabilities, it has become a valuable component to an enterprise’s information management strategy.

3.2 网络附加存储(NAS) Network Attached Storage NAS的组成 NAS的管理 NAS的应用

NAS的发展 NAS连接：文件级别 (File-level)的访问协议，不同于块级别(Block-level)。
Portable Media for File Sharing Networked PCs Networked File Sharing Stand Alone PC NAS连接：文件级别 (File-level)的访问协议，不同于块级别(Block-level)。 In the past, floppy drives with capacities in mere KB’s were widely used to share data files. Over time the need for larger and larger capacity has emerged due to growing need for data to be shared across organizations. Removable storage media, such as flash drives, are capable of storing gigabytes (GB) of data have now complimented the traditional removable media drives. Businesses not only need the capacity to handle huge data storage requirements, the need to share their data has made Network Attached Storage (NAS) an attractive option. NAS systems use external storage for server/hosts, adding flexibility to network storage. NAS works at the file level, rather than the block level. This enables widespread access to the data over the network, based upon the file system client loaded. Network Attached Storage (NAS)

NAS是什么？ NAS是网络上的一套文件级别的共享存储设备。
Clients NAS是一个直接连接到LAN上的、包含高性能文件服务器的存储设备，它专注于提供高速、高效的文件共享服务。 NAS Head Storage NAS is shared storage on a network infrastructure using a unique addressing schema. A NAS server is a storage device that consists of a high performance file server and attached to a LAN. It is a single-purpose machine serving as a dedicated, high-performance, high-speed communication gateway to file data. Note: A NAS device is sometimes called an appliance or filer. The NAS head (as illustrated) could be remote from its storage (gateway) or contained within the same cabinet as its storage—so that the storage is dedicated to NAS applications (integrated). Application Server Print Server NAS Device (Mover/Filer)

通用服务器 vs. NAS From General to Special
Applications Print Drivers File System File System Operating System I/O Network Operating System Network Unlike a general-purpose server, such as a Unix or NT server, a NAS server is a device optimized for file serving functions such as storing, retrieving, and serving files. A single function NAS device provides: Real-time OS dedicated to file serving Open standard protocols Built-in native clustering for high availability Single Function Device (NAS Server: File Serving) General Purpose Server (NT or Unix Server) From General to Special

为什么要有NAS？支持大范围的数据访问改进性能更加灵活：使用标准协议，客户端类型多种多样集中存储简化管理可扩展性
远距离、一对多/多对一、异构平台改进性能特殊的OS、优化的文件服务更加灵活：使用标准协议，客户端类型多种多样 Windows、Unix、Linux、Mac OS 集中存储简化管理可扩展性高可获得性 – clustering 安全保证：用户授权、文件锁 The following are some benefits of NAS: Supports global information access Enables greater file sharing, even over a long distance Supports many-to-one or one-to-many configurations Can share data across platforms Improves efficiency through specialized OS, optimized for file serving Eliminates bottlenecks encountered when accessing files from central file server Relieves general-purpose servers of many file management operations, improving performance of those servers Flexibility - works with many types of clients on both UNIX and Microsoft Windows platforms using Industry standard protocols. Centralizes storage – minimizes duplication on client workstations, reducing management complexity and improving data protection. Simplifies management - leverages existing security infrastructure through standard network protocols. Single point of management for multiple systems for multiple data sets. Identifies data by file name and byte offsets, transfers file data or file meta-data. Scalable - Due to its high performance, low latency design, enables NAS to scale well and depending upon utilization profiles, address many differing types of business applications. High availability Replication and recovery options Can safely centralize large amounts of user data behind a single NAS device with redundant networking equipment to provide maximum connectivity options. Clustering technology for failover in the event of filer failure Handles security, user authentication, and file locking in conjunction with industry standard security schemas.

NAS的逻辑部件网络接口（网卡）网络文件协议 (NFS/CIFS) 优化的OS
NAS Device Network Interface NFS CIFS IP Network NAS Device OS 网络接口（网卡）网络文件协议 (NFS/CIFS) 优化的OS DART - Data Access in Real Time (EMC) Data ONTAP (Network Appliance) 存储接口（磁盘）：SATA, SCSI, or Fibre Channel Storage Interface SCSI, FC, or ATA A NAS device is made up of the following components: Network Interface via one or more Network Interface Cards (NICs) Examples: Gigabit Ethernet (1000 Mb/s), Fast Ethernet (10Mb/s), ATM, and FDDI. Network File Systems (NFS) and Common Internet File Systems (CIFS) protocols Proprietary, optimized Windows, UNIX, or LINUX based OS. Examples: DART - Data Access in Real Time (EMC) Data ONTAP (Network Appliance) Industry standard storage protocols to connect to and manage physical disk storage resources. Examples: Serial ATA (SATA), SCSI, or Fibre Channel

网络文件协议 (NFS/CIFS) NAS Device File-level Network Interface NFS CIFS
Unix NAS Device File-level NFS Network Interface NFS CIFS Redirection IP Network NAS Device OS TCP/IP CIFS Storage Interface Windows Block-level Most NAS devices support multi-protocol file services to handle file I/O requests to the remote file system. The more common protocols for file sharing are: Network File Systems (NFS) - developed by Sun and closely aligned with UNIX-based operating systems Common Internet File Systems (CIFS) – developed by Microsoft and closely aligned with Windows-based operating systems These file system protocols allow users to share file data across different operating environments as well as provide a means for users to transparently migrate from one operating system to another. File system is mounted remotely using NFS or CIFS protocol Application I/O requests transparently transmit data to the remote file system by the NFS/CIFS protocol. This is also known as redirection. Utilizes mature data transport (TCP/IP) and media access protocols NAS device assumes responsibility for organizing block level data (R/W) on disk and managing cache Note: In NAS, I/O from the client is handled by network file access protocols (NFS/CIFS) at the file level, rather than at the block level. Meaning that the client does not need to be aware of physical disk volumes or disk characteristics. Unlike block I/O, there is no awareness of a disk volume, LBA, or disk sector in a file I/O request. The NAS appliance operating system is responsible for keeping track of where files are located on the physical disk subsystem. The NAS device OS issues block I/O requests to the physical disks to fulfill the file I/O read and write requests it receives from clients. SCSI, FC, or ATA

NFS (Network File System)
由Sun开发，与Unix紧密结合一种C/S应用基于TCP，使用RPC (Remote Procedure Calls) 将远程文件系统Mount到本地可以控制对远程Mount的文件系统的访问权限 NFS is a client/server application that enables a computer user view and optionally store and update files on a remote computer as though they were on the user's own computer. It uses Remote Procedure Calls (RPC) to communicate between computers. The user's system requires an NFS client to connect to the NFS server. Since the NFS server and client use TCP/IP to transfer files, TCP/IP must be installed on both systems. Using NFS, the user or system administrator can mount all or a portion of a file system (which is a portion of the hierarchical tree in any file directory and subdirectory). The portion of the file system that is mounted (designated as accessible) can be controlled using permissions (e.g., read-only or read-write).

CIFS (Common Internet File System)
由Microsoft开发，与Windows紧密结合 Microsoft SMB (Server Message Block) 协议的公开版客户端通过SMB协议访问远程文件通常看作是对FTP、HTTP的补充对文件的控制比FTP更好程序接口也比 Web browsers和HTTP更好通过特殊的文件锁共享文件（写）文件名使用Unicode编码 CIFS is client/server application protocol, which enables clients programs make requests for files and services on remote computers on the Internet. CIFS is a public (or open) variation on Microsoft’s Server Message Block (SMB) protocol. SMB is widely used on LANs. Like SMB, CIFS runs at a higher level than, and uses the Internet's TCP/IP protocol. CIFS is viewed as a complement to the existing Internet application protocols such as the File Transfer Protocol (FTP) and the HyperText Transfer Protocol (HTTP). The CIFS protocol allows the client to: Get access to files that are local to the server and read and write to them Share files with other clients using special locks Restore connections automatically in case of network failure Use Unicode file names In general, CIFS gives the client user better control of files than FTP. It provides a potentially more direct interface to server programs than currently available through a Web browser and the HTTP protocol.

NAS连接：A Closer Look OSI Seven-Layer Module Internet Protocol Suite
Application NFS FTP, Telnet SMTP, SNMP Presentation XDR (eXternal Data Representation) Session RPC Transport TCP, UDP Network IP ARP / RARP Data Link While CIFS and NFS are file system protocols, it is important to understand how the network transport protocols of IP, TCP, and FTP fit into the picture. OSI model (developed by the ISO standards body) - defines the specific layers that are responsible for communication tasks. Internet Protocol Suite – defines a group of open-system (non-proprietary) protocols that communicate across interconnected networks (LAN/WAN). This suite includes both low layer protocols (e.g., IP and TCP) as well as common applications such as electronic mail, terminal emulation, and file transfer (FTP). IP is a network-layer protocol that contains addressing information and some control information, enabling packets to be routed. In NAS, the back-end connects to its storage most often using Fibre Channel interconnectivity and the front-end/client connectivity most often via the TCP/IP protocol. If any client wants to access a file from NAS system, it requests the file directly. The NAS system then converts this request in block level access and retrieves data from storage and presents data to client as a complete file. Not Defined Physical OSI Seven-Layer Module Internet Protocol Suite

Block I/O to storage device
Application Storage Interface Block I/O to storage device Operating System Storage Protocol I/O Redirect NAS Operating System NFS / CIFS NFS / CIFS NFS and CIFS protocols handle file I/O requests to the remote file system, which is managed by the NAS device. I/O requests are packaged by the requestor into TCP/IP and forwarded through the network stack, transported across the network, and received by the NAS. The NAS converts the protocol request into an appropriate physical storage request (block I/O), and then performs the operation against the physical storage pool. The data returned from the physical storage pool is then processed by the NAS and repackaged into an appropriate file protocol response. This response is packaged into TCP/IP again and forwarded through the network to the client. This example shows an operation being directed to the remote NAS device and how the different protocols and software layers play a part in moving the request and response between the client and NAS. TCP/IP Stack TCP/IP Stack Network Interface Network Interface Client IP Network NAS Device

UNIX and Windows之间的信息共享
NFS Traffic FTP CIFS Traffic Protocol Layer Common File System (CFS) Multi-protocol support layer OS Due to the structure of the specialized operating system on NAS devices, multiple protocol stacks can be simultaneously supported, thereby allowing disparate systems access to the storage simultaneously. I/O layer

NAS的物理部件服务器（NAS head/Data movers/filers）：高端文件服务器
CPU、Memory, 网络端口、存储连接, … 基于IP的文件服务特殊的 OS 管理接口：进行管理和配置的设备管理NAS Head，配置网络接口配置、管理远程文件系统可以是本地设备或远程设备连接（Connectivity） NAS head to storage / NAS head to network 存储（Storage）集成的存储（Integrated）共享的存储（Gateway，如SAN） NAS devices are comprised of similar components: Data Movers/Filers - Essentially high-end file servers which move data between the network and the storage. Contain memory, ports for network (Ethernet) and storage connectivity, and other common server components. Present file systems to desktop clients and application servers as a shared storage device/file system on an IP network. Uses a specialized OS Management interface - A controlling device for management and configuration Connectivity - The NAS head is connected to the back-end Fibre Channel HBA’s. NAS HBA’s can be connected directly to the storage device or can use Fabric switches. Storage - High performance, dedicated storage configured to provide high level of data availability and data protection using combination of RAID configurations. The storage can be: Integrated (dedicated, directly attached to NAS devices) Gateway (shared, directly or fabric attached devices)

Integrated vs. Gateway NAS
Integrated NAS IP Network NAS Head NAS Gateway There are two types of NAS devices available today: Integrated - combines the NAS controller and the dedicated storage into a single packaged solution. Gateway – the NAS controller is separate and configured to use pre-existing shared external storage such as that found on a SAN. The storage can be connected via FC. FC Fabric IP Network NAS Head

Integrated NAS Integrated NAS System Integrated NAS的优点: 集中的存储介质易于管理
Direct Attach IP Network NAS Head Storage Integrated NAS的优点: 集中的存储介质易于管理易于备份，以保护关键数据高可获得性 In this type of configuration, back-end storage is directly connected to the NAS Head. It is dedicated to one NAS head, not shared with any other servers. Storage is utilized to maximum capacity by the NAS head to get maximum throughput from such a configuration. The benefits of integrated NAS include: Pooled / centralized storage Easier administration Protected mission critical data with backup features (disk & tape) Higher availability Heterogeneous file sharing No need to do storage management Offloads SAN cycles Increased performance throughput (service level) to end users Scalable Lower Total Cost of Ownership 异构系统的文件共享更高的吞吐率可扩展性好更低的TOC（ Total Cost of Ownership）

Gateway NAS SAN NAS Gateway 结合了 NAS 和 SAN 的优点: SAN 的可扩展性及性能
拓展了FC拓扑结构的局限性使得IP设备可以访问SAN 增加了SAN的投资收益降低了访问代价可以更加充分地使用SAN，提高其使用效率支持异构系统的文件服务 Clients IP Network Application Servers IP Network Different types of back-end storage can be attached to one NAS head. SAN In this configuration, back-end storage is shared between different types of application servers (SAN environment) and NAS heads. Different types of back-end storage can be attached to one NAS head. The NAS head boots from one storage and users pool storage for clients from other storage. This is a typical SAN setup, where the NAS head works as gateway to SAN environment. NAS Gateways provide the same benefits and characteristics of NAS: Connects to IP networks Performs as a file server Heterogeneous file sharing Data protection Diagnostic capabilities Clustering and failover features NAS Gateway gives you the combined benefits of NAS and SAN: SAN scalability and performance NAS flexibility and ease of use Increases the reach of SAN infrastructure Extends beyond topology limitations of FC Provides IP device access to SAN storage Leverages the value of SAN investment Reduces access costs. Allows access to underutilized SAN storage. Enables heterogeneous file serving Note: Some NAS Gateway designs offer multiple connectivity options: FC switches, hubs, directors, RAID controllers and disk arrays. FC Switch NAS Gateway Storage

NAS的管理大多数NAS设备提供商都提供了相应的管理软件 NAS管理的话题包括：
备份（Backup）与恢复（Recovery）资源管理（Resource management）：硬件设备及配置、数据空间管理：用户及其使用情况容量管理：存储的容量性能分析：I/O性能 - CPU/Memory、Cache、NFS/CIFS 监控（Monitoring）：CPU、Memory利用率、网络通常需要做预先的分析：目录信息（设备、容量、使用率、性能等）、备份/恢复策略、当前的工具可能还需要一些辅助软件 Now that you have learned about the components in a NAS environment and ways that NAS can be implemented, we will look at the issues involved in managing a NAS environment. - Describe the issues involved in managing a NAS environment - Differentiate between the issues related to managing an integrated system vs. a gateway system Most NAS devices ship with vendor supplied management software, which typically provides a means to configure a NAS device. Select devices even provide basic device monitoring and performance capabilities. NAS devices have several unique management issues, such as: device availability, backup and recovery in addition to traditional storage management issues, such as: resource management, space management, capacity, and performance analysis. Monitoring CPU utilization, memory utilization along with IP traffic are also factors. All are essential to ensure service levels are met. Preliminary analysis - assess the environment. Inventory of all implemented NAS devices (capacity, utilization, performance, location, etc.). Understand the backup/recovery strategies. Identify what tools are currently managing the NAS device and whether they meet your needs. Understand what needs to be accomplished. Once the inventory of the NAS device(s) is complete, the following conclusions should be reached: The number of NAS devices already installed in the organization. The physical configuration of the NAS device (i.e., number of CPUs, amount of memory, IP connections, etc.). The logical or resource management aspects of the device (mapping of the logical volumes to the physical spindles, mapping of the file systems to the logical volumes and perhaps the allocation attributes of the file systems). The backup/recovery methodologies. The following management concerns need to be considered when using NAS: Performance - includes physical device performance (CPU utilization, memory utilization, number of I/O requests received, cache performance, etc.) as well as CIFS/NFS performance. Management software should not only collect this data, but also filter it so intelligent events can be initiated when thresholds are breached. Discovery - the ability to discover a NAS device(s) –along with its hardware configuration and logical storage attributes—as they change, without manual entry. Space Management - administrators need to know who is using it, what they are using, and how much of it. Backup/Recovery - Backup and recovery for NAS is similar to other file servers. Logical backups at the file and file system level need to be considered along with more database-centric backup and recovery solutions involving data associated with a relational database management system (RDBMS). Many tools are available that cover the full range of backup and recovery options. Asset Management -There are two aspects to asset management: Physical attributes of a NAS device, and its associated configuration Logical or data aspects. Both need to be inventoried and, in most cases, have some level of cost recovery. The ever-changing capacity information of a NAS device needs to be collected in order to correlate that data to utilization. This can be either at the user or application level, with the data being fed into the fixed asset repository to create cost recovery reports.

NAS管理：Integrated vs. Gateway
Integrated NAS的管理 NAS服务器（NAS Head）和存储阵列（Storage Array）都由 NAS管理软件来管理 Gateway NAS的管理 NAS服务器（NAS Head）由 NAS管理软件管理存储阵列（Storage Array）则由其自身的阵列管理软件管理 Managing an Integrated System Since the storage array is dedicated to the NAS functionality, the NAS management software is responsible for the managing both the NAS components and the backend storage array. Managing a Gateway System Gateway NAS Systems use shared storage, which means that traditional SAN hosts can also utilize the same array. Therefore both the NAS hardware and the array are individually managed by their own specialized management software: The NAS component is managed via specialized NAS management software The storage array is managed via its native array management software Supports standard host connectivity where integrated does not (enabling the array to be multipurpose).

NAS示例 NAS存储融合采用 Gateway NAS 节约投资

General purpose OS serving files via FTP, CIFS, NFS, HTTP. . .
NAS存储融合：场景 Current Environment UNIX NT W2K NFS CIFS Internet/Intranet In many companies the need for two differing environments maintains the separation of two technologies using the same infrastructure for the same purpose, but indifferent ways. Access to networked files for UNIX (NFS) and Microsoft (CIFS) are traditionally housed on separate servers infrastructures. UNIX Windows General purpose OS serving files via FTP, CIFS, NFS, HTTP. . .

Special OS serving files via FTP, CIFS, NFS, HTTP. . .
NAS存储融合：解决方案 Solution NAS File Server Internet/Intranet However by implementation of NAS these same file structures can be housed together, while still maintaining their integrity. Within NAS deployments the same file system can be accessed by the same user via different technologies, either NFS or CIFS, and still maintain the integrity of the data and security structures, as long as the applications used for both methodologies understand the data structures presented. Benefits of this solution: Provides continuous availability to files Heterogeneous file sharing Reduces cost for additional OS dependent servers Adds storage capacity non-disruptively Consolidates storage management Lowers Total Cost of Ownership UNIX Windows Special OS serving files via FTP, CIFS, NFS, HTTP. . .

采用 Gateway NAS 节约投资当拥有SAN时，无需再购买Integrated NAS。
Multipurpose Servers NT UNIX IP Network Benefits: Provides continuous availability to files Heterogeneous file sharing Reduces cost for additional OS dependent servers Adds storage capacity non-disruptively Consolidates storage management Lowers Total Cost of Ownership Note: FC has distance limitations and is costly to deploy when compared to cost of departmental or workgroup class servers. FC Switch NAS Gateway

NAS面临的挑战速度可靠性：存在单点失效的问题连接：文件级别（File-level） Block-level NAS
网络延时与网络拥塞 IP网络的协议栈效率低应用对I/O响应的需求总是不断增加可靠性：存在单点失效的问题连接：文件级别（File-level） Block-level NAS 可扩展性：总是会有一个最大的容量，不能无限扩大 Speed Network latency and congestion Protocol stack inefficiency encapsulation, possessor overhead, and relatively small payload Application response requirements Reliability Due to the large geographical coverage of enterprise networks there are inherent possibilities for network failures, but with redundancy planning these issues can be minimized. Centralized storage silos may become single points of failure without remote mirroring or backup facilities. Connectivity Without newly emerging technologies, iSCSI, FCIP & iFCP, many applications required block level access therefore excluding NAS as a solution for businesses Scalability Although NAS devices can scale to terabytes of storage capacity, once the capacity is exhausted the only way to expand is to add additional devices. This can cause additional problems when data center real estate is at a premium Once a NAS device is fully populated, including external storage enclosures, the only remaining scaling option is to buy another system. When data center real estate is at a premium this can be seen as a major limitation.

小结 NAS设备：专注于提供优化的文件服务的设施 NAS的管理 NAS示例：NAS存储融合、Gateway NAS
NAS Head / Storage Array NAS拥有专门的OS NAS 支持多种网络文件协议 NAS的实现方式：Integrated / gateway NAS的管理性能、资源、空间、容量、备份/恢复、监控等 Integrated / gateway：不同的管理方式 NAS示例：NAS存储融合、Gateway NAS NAS面临的挑战：速度、可靠性、连接、可扩展性

本章目标及内容面对不断增长的信息需求和不断发展的网络技术，网络存储应运产生。本章分别介绍了网络存储环境下的直连存储、网络附加存储、光纤存储局域网、IP存储局域网和内容寻址存储，并深入介绍了这些网络存储结构的概念、结构特点、基本原理以及应用环境等方面的知识。本章内容包括5个方面： 3.1 直连存储 (Direct Attached Storage，DAS) 3.2 网络附加存储 (Network Attached Storage，NAS) 3.3 光纤存储局域网络 (Fibre Channel Storage Area Networks, SAN) 3.4 IP存储局域网（IP Storage Area Networks，IP SAN） 3.5 基于对象的存储（Object-based and Storage, OBD） In the previous section, we looked at the components of a storage system. In this section, the emphasis will be on different storage system models-- starting with a basic storage model (DAS) and focusing on networked storage models such as SAN, NAS, and CAS. When effectively designed and implemented, Networked Storage enables more effective utilization of available storage, increased flexibility, and better availability of data. Direct Attached Storage (DAS) - storage devices directly connected host either via an internal or external connection. Network Attached Storage (NAS) - storage devices accessed over a network, in front of the server. Is made up of computing/storage devices running an operating system that has been optimized for file service. Storage Area Network (SAN) - a dedicated network that carries data between computer systems and storage devices. Content Addressed Storage (CAS) – an object-oriented, location independent approach to data storage. It can be implemented using a number of different technologies.

3.3 光纤存储局域网络 (SAN) Fibre Channel Storage Area Network – FC SAN
NAS vs. SAN Direct Attached Storage, or DAS, is the most basic level of storage. DAS devices either reside as an integrated part of the host computer (e.g., hard drives, removable storage devices, etc.) or directly connected to a single server externally (such as RAID arrays or removable media). As the first widely popular storage model, DAS products still comprise a large share of the installed base of storage systems in today's IT infrastructures. In this module, we will look at what DAS is, its components, management issues, and the challenges that DAS presents.

FC SAN概述业务需求及技术挑战用户对信息的需求业务的集成灵活、可扩展的体系架构 On-line storage
一个组织内拥有成千上万的服务器关键数据并不一定在数据中心 24x7 业务的集成孤岛与烟囱安全的操作环境灵活、可扩展的体系架构对业务需求的快速响应减少信息管理的成本 There are many challenges for data center managers who are supporting the business needs of the users such as: Providing information when and where the business user needs it. Things that impact this challenge include: Explosion in on-line storage Thousands of servers through out organization Mission-critical data is not just in the data center 24x7 availability is a requirement Integrating technology infrastructure with business processes to: Eliminate stovepiped application environments Secure operational environments Providing a flexible, resilient architecture that: Responds quickly to business requirements Reduces the cost of managing information

SAN是什么? A dedicated network that carries data between computer systems and storage devices, which can include tape and disk resources. Organized connections among: Storage Communication devices Computer Systems 安全（Secure）健壮（Robust） Array Switches Server A Storage Area Network (SAN) is a dedicated network that carries data between computer systems and storage devices, which can include tape and disk resources. A SAN consists of a communication infrastructure, which provides physical connections, and a management layer, which organizes the connections, storage elements, and computer systems so that data transfer is secure and robust. Servers Storage

来自SNIA的定义存储网络工业协会(Storage Networking Industry Association, SNIA)的定义：“存储区域网络（SAN）是一个高性能的专用网络，它的基本目的是使存储设备同计算机系统或存储设备与存储设备之间通信” SAN的特殊之处：能够进行全局范围的存储连接 SAN的关键之处：它把很多计算机和存储设备连接起来，使这些计算机能够相互协商对设备的所有权在更优秀的SAN中，计算机间能够通过SAN进行数据共享

LAN vs. SAN SAN: A second network - Fabric

Network and Channel Channel: Connectivity among pre-addressed devices Network: Connectivity via protocol

FC SAN的发展 Interconnected SANs SAN Islands FC仲裁环(Arbitrated Loop)
Servers HUB Switches Switches Servers Storage Servers Storage As business demand for data grew, DAS and NAS implementations allowed companies to store and access data effectively, but often inefficiently. Storage was isolated to the specific devices, making it difficult to manage and share. The effort to regain control over the dispersed assets caused the emergence of storage area networks (SANs). SANs had the advantage of centralization, resulting in improved efficiencies. The first implementation of SAN was a simple grouping of hosts and associated storage in a single network, often using a hub as the connectivity device. This configuration is called Fibre Channel Arbitrated Loop (FCAL). It could also be referred to as a SAN Island due to the fact that a) there is limited connectivity and b) there is still a degree of isolation. As demand increased and technology improved, Fibre channel switches replaced hubs. Switches greatly increased connectivity and performance allowing for interconnected SANs and ultimately enterprise level data accessibility of SAN applications and accessibility. SAN Islands FC仲裁环(Arbitrated Loop) Interconnected SANs FC交换光纤(Switched Fabric) Arrays Storage Enterprise SANs FC Switched Fabric

FC SAN的优点高带宽（High bandwidth）扩展的SCSI 资源融合（Resource Consolidation）
Fibre Channel Protocol, 400MB/s 扩展的SCSI SCSI over FC, Block I/O Greater distances still with SCSI functions 资源融合（Resource Consolidation） Centralized storage and management 可扩展性（Scalability） Up to 16 million devices 安全访问（Secure Access） Isolation and filtering Some of the benefits of implementing a SAN are discussed here. A SAN uses the Fibre channel transport which is a set of standards which define protocols for performing high speed serial data transfer, up to 400 Megabytes per second. It provides a standard data transport medium over which computer systems communicate with devices such as disk storage arrays. SCSI over Fibre Channel implementations allow these devices to be connected in dynamic Fibre Channel topologies which span much greater distances and provide a greater level of flexibility and manageability while retaining the basic functionality of SCSI. Fibre Channel networks are often referred to as networks that perform channel operations. As it is a networked infrastructure, many devices and host can be attached seamlessly, upwards of 16 million devices in a SAN. This allows better utilization of corporate assets and ease of management both for configuration and security.

FC SAN的组成 Host Bus Adapter (HBA) Fiber Cabling
including the device drivers needed to communicate within the SAN Fiber Cabling optical or copper Fibre Channel Switch /Hub devices used to connect the nodes Storage Array Management System HBA Switches HBA SAN-attached Server As can be seen from the graphic on this page, a SAN consists of three basic components – server(s), the SAN infrastructure and the storage. Each of these components can be broken down into even more finite components such as: A Host Bus Adapter (HBA) which is installed in a server (including the device drivers needed to communicate within the SAN). Cabling which is usually optical but can be optical or copper. Fibre Channel switches or hubs – devices used to connect the nodes. Storage arrays. Management system to analyze and configure SAN components. SAN Arrays

Nodes, Ports, & Links Link Port 0 Rx Tx Port 0 Port 1 A node can be considered any device that is connected to the SAN for purposes of requesting or supplying data (e.g. servers and storage). Nodes use ports to connect to the SAN and to transmit data. There are two connection points on a port, a transmit (Tx) link and a receive (Rx) link. Data traveling simultaneously through these links is referred to as Full Duplex. A node can be considered any device that is connected to the SAN for purposes of requesting or supplying data (e.g. servers and storage). Port n Node

HBA：Host Bus Adapters Port Node 位于主机的I/O总线和光纤通道网络之间。
HBAs可以自动执行低级别的接口操作以降低对主机处理器的影响。类似于LAN中的网卡（NIC）。应用与存储之间关键的一环。 Port Switches Arrays The Hosts connect to the SAN via an HBA. As referenced on the previous slide, the host would be the node and the HBA would represent the port(s). HBAs can be compared to a NIC in a Local Area Network, as they provide a critical link between the SAN and the operating system and application software. An HBA: Sits between the host computer's I/O bus and a Fibre Channel network and manages the transfer of information between the two channels. Performs many low-level interface functions automatically or with minimal processor involvement, such as I/O processing and physical connectivity between a server and storage. Thus, the HBAs provide critical server CPU off-load, freeing servers to perform application processing. As the only part of a storage area network that resides in a server, HBAs also provide a critical link between the SAN and the operating system and application software. HBA HBA Server Node

FC SAN的连接（Connectivity）
多模光纤(Multimode fiber, MMF): multiple light rays 规格：50，62.5微米(micron) Short distance: 2~500m; 可通过模式色散(modal dispersion)技术改进 Nodes to switch (hub) 单模光纤(Single mode fiber, SMF): a single light ray 7~11 microns ( 9 microns) Long distance: 2m~10km Switch to switch 铜缆线(Copper) 0~24m To connect the nodes, optical fiber cables are used. There are two types of cable employed in a SAN – Multimode and Single mode. Multimode fiber (MMF) can carry multiple light rays, or modes, simultaneously. MMF typically comes in two diameters – 50 micron and 62.5 micron ( a micron is a unit of measure equal to one millionth of a meter). MMF transmission is used for relatively short distances because the light tends to degrade, through a process called modal dispersion(模式色散), over greater distance. MMF is typically used to connect nodes to switches or hubs. For longer distances, single mode fiber (SMF) fiber is used. It has a diameter of 7 – 11 microns with 9 microns being the most common and transmits a single ray of light as a carrier. As there is less bounce, the light does not disperse as easily, allowing long-distance signal transmission. This type of cable is used to connect two switches together in a SAN. Single Mode Fiber Switches Storage Multimode Fiber Host

连接器：Connectors 节点连接器（Node Connectors）光纤配线盘连接器（Patch panel Connectors）
SC Duplex Connectors for 1Gb fiber optic cable LC Duplex Connectors for 2Gb or 4Gb fiber optic cable 光纤配线盘连接器（Patch panel Connectors） ST Simplex Connectors the first standard for fiber optic cabling Optical and Electrical connectors are used in SANs. The SC connector is the standard connector for fiber optic cables used for 1Gb. The LC connector is the standard connector for fiber optic cables used for 2Gb or 4 Gb. The ST connector is a fiber optic connector which uses a plug and socket which is locked in place with a half-twist bayonet lock. Often used with Fibre Channel Patch Panels. (Note: A Patch Panel(光纤配线盘 ) is generally used for connectivity consolidation in a data center.) The ST connector was the first standard for fiber optic cabling.

连接设备 SAN通信的基础 Hubs: FCAL, 所有节点共享带宽
Switches and Directors: more “intelligent” devices 点对点直接通信 HBA For Fibre Channel SANs, connectivity is provided by Fibre Channel hubs, switches and directors. These devices act as the common link between nodes within the SAN. Connectivity devices can be categorized as either hubs or switches. A hub is an communications device is used in FCAL and which physically connects nodes in a logical loop/physical star topology. This means that all nodes must share bandwidth , as data travels through all connection points. A Fibre Channel switch/ director is a more “intelligent” device. It has advanced services that can route data from one physical port to another directly. Therefore each node has a dedicated communication path, aggregating bandwidth in the process. Compared to switches, directors are larger devices deployed for data center implementations. They function similarly to switches but have higher connectivity capacity and fault-tolerant hardware. Arrays Switches Server

存储资源 Storage Array Features of an array
Provides storage consolidation and centralization Features of an array High Availability/Redundancy Performance Business Continuity Multiple host connect Arrays Switches The fundamental purpose of any SAN is to provide access to storage – typically storage arrays. As discussed previously, storage arrays support many of the features required in a SAN such as: High Availability/Redundancy Improved Performance Business Continuity Multiple host connect HBA Server

SAN管理软件一套用于管理SAN中主机和存储阵列之间接口的工具提供对SAN环境的集成管理 Web based GUI or CLI
在服务器、交换机、存储设备之间建立映射监控和管理发现的存储设备对SAN进行逻辑划分 SAN管理系统是一台服务器或一个控制台，SAN中的对象可以被监控和维护。它可以通过集中的方式对SAN进行全局管理，以降低复杂性。 SAN Management Software provides a single view of your storage environment. Management of the resources from one central console is simpler and more efficient. SAN management software provides core functionality, including: Mapping of storage devices, switches, and servers Monitoring and alerting for discovered devices Logical partitioning of the SAN Additionally, it provides management of typical SAN components such as HBAs, storage devices and switches. The management system of a SAN is a server, or console, where the objects in the SAN can be monitored and maintained. It offers a central location for a full view of the SAN, thereby reducing complexity.

FC SAN的连接 FC SAN 的连接方法及其拓扑结构光纤通道的设备光纤通道的通信协议光纤通道的登录过程

Servers & Applications
FC SAN Connectivity Core networking principles applied to storage Networked data system, Similar to LAN Servers are attached to 2 distinct networks Back-end Front-end Back-end Network IP network SAN switches directors Front-end Network SANs combine the basic functionality of storage devices and networks, consisting of hardware and software, to obtain a highly reliable, high-performance, networked data system. Services similar to those in any LAN (e.g. name resolution, address assignment etc.) allow data to traverse connections and be provided to end-users. When looking at an overall IT infrastructure, the SAN and LAN are mutually exclusive but serve similar purposes. The LAN allows clients, such as desktop work-stations, to request data from servers. This could be considered the front-end network. This is where the average user would connect typically across an Ethernet network. The SAN, or back-end network also connects to servers, but in this case, the servers are acting as clients. They are requesting data from their servers – the storage arrays. These connections are accomplished via a Fibre Channel network. (Note: FibRE refers to the protocol versus fibER which refers to a media!) By combining the two networks together, with the servers as the common thread, the end-user is supplied with any data they may need. Users & Application Clients Servers & Applications Storage & Application Data

光纤通道（Fibre Channel）是什么？
Fibre Channel Protocol (FCP) FC-0（物理）, FC-1（编解码）, FC-2（编址）, FC-3（尚未实现）, FC-4（应用）—— Similar to OSI Model 24-bits addressing scheme – 224 (16 million) devices in a single SAN SAN Transport Protocol Integrated set of standards (ANSI) Encapsulates SCSI A High Speed Serial Interface Allows SCSI commands to be transferred over a storage network. Standard allows for multiple protocols over a single interface. Fibre channel is a set of standards which define protocols for performing high speed serial data transfer. The standards define a layered model similar to the OSI model found in traditional networking technology. Fibre Channel provides a standard data transport frame into which multiple protocol types can be encapsulated. The addressing scheme used in Fibre Channel switched fabrics will support over 16 million devices in a single fabric. Fibre Channel has become widely used to provide a serial transport medium over which computer systems communicate with devices such as disk storage arrays. These devices have traditionally been attached to systems over more traditional channel technologies such as SCSI. SCSI over Fibre Channel implementations now allow these devices to be connected in dynamic Fibre Channel topologies which span much greater distances and provide a greater level of flexibility and manageability than found with SCSI. Fibre Channel networks are often referred to a networks that perform channel operations.

光纤通道的端口（Port） G_Port - either a F_Port or a E_port Node Node Node Node
Servers NL Port NL Port NL Port Server Node Node HUB HUB Storage NL Port NL Port Node FL Port FL Port Server E Port F Port N Port Switch Fibre Channel ports are configured for specific applications Host Bus Adapters and Symmetrix FC Director ports are configured as either N-Ports or NL-Ports N_Port - Node port, a port at the end of a point-to-point link NL_Port - A port which supports the arbitrated loop topology Fibre Channel Switch ports are also configured for specific applications F_Port - Fabric port, the access point of the fabric which connects to a N_Port FL_Port - A fabric port which connects to a NL_Port E_Port - Expansion port on a switch. Links multiple switches G_Port - A switch port with the ability to function as either a F_Port or a E_port Note: Port Type is defined by Firmware / HBA Device Driver configuration Settings. E Port F Port Node Switch N Port Node F Port F Port Switch Node N Port N Port Node Array Array Storage

WWN：Unique 64 bit identifier.
FC设备的World Wide Names WWN：Unique 64 bit identifier. Static to the port. Used to physically identify a port or node within the SAN Similar to a NIC MAC address Additionally, each node is assigned a unique port ID (24 bit address) within the SAN Used to communicate between nodes within the SAN Similar in functionality to an IP address on a NIC All Fibre Channel devices (ports) have 64 bit unique identifiers called World Wide Names (WWN). These WWNs are similar to the MAC address used on a TCP/IP adapter, in that they uniquely identify a device on the network and are burned into the hardware or assigned through software. It is a critical feature, as it used in several configurations used for storage access. However, in order to communicate in the SAN, a port also needs an address. This address is used to transmit data through the SAN from source node to destination node.

World Wide Names示例 5 6 1 B 2 1 c 9 2 d 4 World Wide Name – Array
6 1 B 2 0101 0000 0110 0001 1011 0010 Company ID 24 bits Port Model seed 32 bits World Wide Name - HBA 1 c 9 2 d 4 Reserved 12 bits Company OUI 24 bits Company Specific 24 bits In this example: When a N_Port is connected to a SAN, an address is dynamically assigned to the port. The N_Port then goes through a login at which time it registers its WWN with the Name Server. Now the address is associated with the WWN. If the N_Port is moved to a different port on the fabric, its address will change. However, the login process is repeated so the WWN will become associated with the new N_Port address. This allows for configuration to take advantage of the fact that the WWN has remained the same even though the FC address has changed

Port ID示例 When a N_Port is connected to a SAN, an address (port ID) is dynamically assigned to the port. The N_Port then goes through a login at which time it registers its WWN with the Name Server. 当N_Port移动，port ID发生变化，需要重新Login。域ID 16 15 分区ID 8 7 端口ID 0 N Port ID 环ID 7 AL-PA ID 0 Public N Port ID unused 7 AL-PA ID 0 Private NL Port ID

SAN中的设备必须向存储网络登录 (login) 之后才能进行通信。SAN的光纤通道中有三种类型的登录：
Fabric, Port, Process Fabric N Port 1 F Port F Port N Port 2 Process a Process x In order for a device to communicate on the SAN it must authenticate or login to the storage network. There are three types of login supported in Fibre Channel: Fabric – All node ports must attempt to log in with the Fabric (A Fabric is the ‘complete’ SAN environment.) This is typically done right after the link or the Loop has been initialized. Port – Before a node port can communicate with another node port, it must first perform N_Port Login with that node port. Process – Sets up the environment between related processes on node ports. By completing this login process, nodes have the ability to transmit and receive data. Process b Process y Process c Process z

光纤通道的寻址（Addressing） Fibre Channel addresses (port ID) are used for transporting frames from source ports to destination ports. 地址（port ID）分配方法：不同的拓扑结构有不同的分配方法 (loop vs. switch) Loop – self assigning Switch – centralized authority 保留(reserved)的地址 FFFFFA is Management Server FFFFFC is Name Server FFFFFE is Fabric Login Server FFFFFF is Broadcast Address …… As mentioned previously, FC addresses are required for node communication. Fibre Channel addresses are used to designate the source and destination of frames in the Fibre Channel network. These addresses could be compared to network IP addresses. They are assigned when the node either enters the loop or is connected to the switch. There are reserved addresses, which are used for services rather than interface addresses. IP network broadcast addresses are ranged from to

Fabric是什么？ Virtual space used by nodes to communicate with each other once they are joined. Component identifiers: Domain ID: Switch World Wide Name (WWN): Node Fabric services: Login service Name service Fabric controller Management server …… Arrays Fabric Servers Switches A fabric is a virtual space in which all storage nodes communicate with each other over distances. It can be created with a single switch or a group of switches connected together. Each switch contains a unique domain identifier which is used in the address schema of the fabric. In order to identify the nodes in a fabric, 24-bit fibre channel addressing is used. Fabric services: When a device logs into a fabric, its information is maintained in a database. The common services found in a fabric are: Login Service Name Service Fabric Controller Management Server Storage

光纤通道的拓扑结构（ Topologies）
仲裁环(Arbitrated Loop, FC-AL) 设备与一个共享的环(loop)连接类似于令牌环 (Token Ring) 8-bits地址交换光纤(Switched Fabric, FC-SW) 所有设备与 “Fabric Switch”连接类似于 IP switch 发起者 (Initiators)到目标 (Targets)之间有唯一的、专门的I/O路径 24-bits地址 HUB Clients Storage Arrays The ANSI Fibre Channel Standard defines distinct topologies: Arbitrated loop (FC-AL), Switched fabric (FC-SW). Arbitrated loop (FC-AL) - Devices are attached to a shared “loop”. FC-AL is analogous to the token ring topology. Each device has to contend for performing I/O on the loop by a process called “arbitration” and at a given time only one device can “own” the I/O on the loop - resulting in a shared bandwidth environment. Switched Fabric - Each device has a unique dedicated I/O path to the device it is communicating with. This is accomplished by implementing a fabric switch. Switch Clients

Switch vs. Hub Switches (FC-SW) Hubs (FC-AL)
FC-SW architecture scalable to millions of connections. Bandwidth per device stays constant with increased connectivity. Bandwidth is scalable due to dedicated connections. Higher availability than hubs. Higher cost. Hubs (FC-AL) FC-AL is limited to 127 connections (substantially fewer connections can be implemented for ideal system performance). Bandwidth per device diminishes with increased connectivity due to sharing of connections. Low cost connection. The primary differences between switches and hubs are scalability and performance. The FC-SW architecture scales to support over 16 million devices. Expansion ports, explained within the next few pages, must be implemented on switches to allow them to interconnect and build large fabrics. The FC-AL protocol implemented in hubs supports a maximum of 126 nodes. As discussed earlier, fabric switches provide full bandwidth between multiple pairs of ports in a fabric. This results in a scalable architecture which can support multiple communications at the same time. The hub on the other hand provides shared bandwidth which can support only a single communication at a time. Hubs provide a low cost connectivity expansion solution. Switches, on the other hand, can be used to build dynamic, high-performance fabrics through which multiple communications can occur at one time and are more costly.

仲裁环集线器（ Hub）的工作原理 Node A Node D Node B Node C Byp Byp Byp Byp Transmit
NL_Port #1 HBA NL_Port #1 HBA Hub_Pt Hub_Pt NL_Port #2 HBA NL_Port #2 HBA Transmit Receive Byp Byp Byp Byp Receive Transmit Node B Node C NL_Port #4 HBA NL_Port #4 HBA Transmit Receive NL_Port #3 FA NL_Port #3 FA Byp Byp Byp Byp Receive Transmit Hub_Pt Hub_Pt

交换光纤（ Switched Fabric）的工作原理
A node contacts Name Service to obtain list of nodes already logged in. Node A Node D NL_Port #1 HBA Port Port NL_Port #2 HBA Transmit Receive N_Port #2 Storage Port N_Port #1 HBA Receive Transmit Node B Node C FCSW: At boot time, a node initializes and logs into the fabric. The node contacts Name Service to obtain list of nodes already logged in. Node attempts individual device logins and transmits data via the FCSW. This link is considered a dedicated connection between the initiator and the target. All subsequent exchanges between these nodes will make use of this "private" link. NL_Port #4 HBA N_Port #4 HBA Transmit Receive NL_Port #3 FA N_Port #3 Storage Port Receive Transmit Port Port

Inter Switch Links (ISLs)
E_Port E_Port Multimode Fiber 1Gb=500m 2Gb=300m Switch Switch Single-mode Fiber up to10 km Switch Switch Switches are connected to each other in a fabric using Inter-switch Links (ISL). This is accomplished by connecting them to each other through an expansion port on the switch (E_Port). ISLs are used to transfer host-to-storage data, as well as fabric management traffic, from one switch to another and, hence, they are the fundamental building blocks used in shaping the performance and availability characteristics of a fabric and the SAN. Metro ring or point-to-point topologies with or without path protection Switch Switch Router Router

拓扑结构: Mesh Fabric 两种不同的结构：different hops 所有的switch互相连接
partial mesh or full mesh 所有的switch互相连接主机和存储设备可以在“Fabric”的任意地方主机和存储设备也可以同时连在一个switch上 In this topology, all switches are connected to each other directly using ISLs. The purpose of this topology to promote increased connectivity within the SAN – the more ports that exists, the more nodes that can participate and communicate. Features of a partial mesh topology: Traffic may need to traverse several ISLs (hop) Host and storage can be located anywhere in the fabric. Host and storage can be localized to a single director or switch. Features of a full mesh topology: Maximum of one ISL link or hop for host to storage traffic. Partial Mesh Full Mesh

主机和存储设备可以在“Fabric”的任意地方所有的主机/存储设备之间最多只需1跳（ISL hop）
Full Mesh的优点主机和存储设备可以在“Fabric”的任意地方所有的主机/存储设备之间最多只需1跳（ISL hop）当ISL出现问题时，采用 Fabric Shortest Path First (FSPF) 算法来寻找新的I/O路径 Fabric 的管理更简单 When implementing a mesh topology, follow these recommendations: Localize hosts and storage when possible - Remember traffic will be bi-directional for both read/write and host/storage on both switches. Evenly distribute access across ISLs. Attempt to minimize hops - Traffic from remote switches should represent no more than 50% of overall traffic locally. Fabric Shortest Path First (FSPF) is a protocol used for routing in Fibre Channel switched networks. It calculates the best path between switches, establishes routes across the fabric and calculates alternate routes in event of a failure or topology change. There are some tradeoffs to keep in mind when implementing mesh fabrics such as: Additional switches raise ISL port count and reduce user port count. Thought must be given to the placement of hosts/storage or ISLs can become overloaded or underutilized.

拓扑结构: Simple Core-Edge Fabric
可以是2层或3层结构 Single Core Tier One or two Edge Tiers 在2层结构中，存储设备与 Core Tier连接优点 High Availability Medium Scalability Medium to maximum Connectivity Host Tier Storage Tier In this topology, several switches are connected in a “Hub and Spoke” configuration. It is called this as there is a central connection much like the wheel of a bicycle (Note: This DOES NOT refer to an FCAL hub, it is simply descriptive). There are two types of switch tiers in the fabric: Edge Tier Usually departmental switches. Offers an inexpensive approach to adding more hosts into the fabric. Fans out from the Core tier. Nodes on the edge tier can communicate with each other using the Core tier only. Host to Storage Traffic has to traverse a single ISL (two-tier) or two ISLs (three-tier). Core or Backbone Tier Usually Enterprise Directors. Ensures highest availability since all traffic has to either traverse through or terminate at this tier. With two-tier, all storage devices are connected to the core tier, facilitating fan-out. Any hosts used for mission critical applications can be connected directly to the storage tier, thereby avoiding ISLs for I/O activity from those hosts. This topology increases connectivity within the SAN while conserving overall port utilization. General connectivity is provided by the “core” while nodes will connect to the “edge”. If expansion is required, an additional edge switch can be connected to the core. This topology can have two variations: Two-tier topology (one Edge and one Core as shown) – All hosts are connected to the edge tier and all storage is connected to the core tier. Three-tier topology (two Edge and one Core) – All hosts are connected to one edge; all storage is connected to the other edge; and the core tier is only used for ISLs. Edge Tier Switches Director Core Tier Core Tier Director Switches Edge Tier Switches Edge Tier

Efficient design based on node type.
Core-Edge结构的优点简化“Fabric”的数据传播 One ISL hop access to all storage in the fabric. Efficient design based on node type. Traffic management and predictability. Easier calculation of ISL loading and traffic patterns A key benefit of the core/edge topology is the simplification of fabric propagation. Configurations are easily distributed throughout the fabric due to the common connectivity. Node workloads can be evenly distributed based on location—hosts on the edge, storage in the core. Performance analysis and traffic management is simplified since load can be predicted based on where each node resides. Increasing number of core switches grows ISL count. This is assumed to be a natural progression when growing the fabric but may cause additional hops, thus decreasing performance. Choosing the wrong switch for the core makes scaling difficult. High port-density directors are best suited at the core.

FC SAN的管理基础设施的保护（Infrastructure protection）
Fabric管理（Fabric Management）存储分配（Storage Allocation）容量管理（Capacity Tracking）性能管理（Performance Management） There are several ways to look at managing a SAN environment; Infrastructure protection - One crucial aspect of SAN management is environmental protection, or security. In order to ensure data integrity, steps must be performed to secure data and prevent unauthorized access. This includes physical security (physical access to components) and network security. Fabric Management - Monitoring and managing the switches is a daily activity for most SAN administrators. Activities include accessing the specific management software for monitoring purposes and zoning. Storage Allocation - This process involves making sure the nodes are accessing the correct storage in the SAN. The major activity is executing appropriate LUN Masking and mapping utilities. Capacity Tracking - Knowing the current state of the storage environment is important for proper allocation. This process involves record management, performance analysis and planning. Performance Management - Applications must function equal to, if not better than a DAS environment. Performance Management assists in meeting this requirement as it allows the SAN admin to be aware of current environmental operations, as well as to avoid any potential bottlenecks.

Management LAN (Private)
基础设施的保护 Corporate LAN 物理安全数据中心上锁集中的服务器和存储设施访问控制 Secure VPN or Firewall Servers Management LAN (Private) Control Station Switch Switch It is imperative to maintain a secure location and network infrastructure. The continuing expansion of the storage network exposes data center resources and the storage infrastructure to new vulnerabilities. Data aggregation increases the impact of a security breach. Fibre Channel storage networking potentially exposes storage resources to traditional network vulnerabilities. For example, it is important to: Ensure that the management network, typically IP based is protected via a firewall Password are strong Completely isolate the physical infrastructure of the SAN In-band (FC) Out-band (IP) Storage Arrays

Switch/Fabric 管理工具设备提供商提供的管理软件功能第三方的管理软件 Embedded within the switch
Graphical User Interface (GUI) or Command Line Interface (CLI) 功能 Common functions Performance monitoring Discovery Access Management (Zoning) Different “look and feel” between vendors 第三方的管理软件 Switch vendors embed their own management software on each of their devices. By connecting to the switch across the IP network, an administrator can access a graphical management tool (generally web-based) or issue CLI commands (via a telnet session). Once connected, the tasks are similar across vendors. The difference lies in the commands that are executed and the GUI. Some of the management activities include: Switch Hardware monitoring – ports, fans, power-supplies Fabric activity – node logins, data flow, transmission errors Fabric partitioning – creating managing and activating zones In addition to vendor specific software tools, there are newer SAN management packages being developed by third parties, such as Storage Resource Management (SRM) software. This software monitors a SAN and, based on policies, automatically performs administrative tasks.

Too many server nodes and storage nodes.
Fabric的管理：Zoning Zoning is是一种交换技术，它可以将Fabric中的一些节点从逻辑上划分为可以相互通信的组（Zone）。Zoning模块在login时就让在同一个 Zone中的端口建立链接（Link）。 Servers Zoning is a switch function that allows nodes within the fabric to be logically segmented into groups that can communicate with each other. The zoning function controls this process at login by only letting ports in the same zone to establish link level services. Arrays Too many server nodes and storage nodes.

Zoning的层次及组成在同一时刻，每个Fabric中只有一个Zone set 是活动的。 Zone Set Zone
Zones Sets (Library) Zone Zones (Library) Members (WWN’s) Member A port or a node can be members of multiple zones. There are several configuration layers involved in granting nodes the ability to communicate with each other: Members - Nodes within the SAN which can be included in a zone. Zones - Contains a set of members that can access each other. A port or a node can be members of multiple zones. Zone Sets - A group of zones that can be activated or deactivated as a single entity in either a single unit or a multi-unit fabric. Only one zone set can be active at one time per fabric. Can also be referred to as a Zone Configuration.

Zoning的类型 Servers Domain ID = 21 Port = 1 WWN 10:00:00:00:C9:20:DE:56 Array Switches Domain ID = 25 Port = 3 WWN 10:00:00:00:C9:20:DC:40 In general, zoning can be divided into three categories: WWN zoning (Soft) - WWN zoning uses the unique identifiers of a node which have been recorded in the switches to either allow or block access. A major advantage of WWN zoning is flexibility. The SAN can be re-cabled without having to reconfigure the zone information since the WWN is static to the port. Port zoning (Hard) - Port zoning uses physical ports to define zones. Access to data is determined by what physical port a node is connected to. Although this method is quite secure, should re-cabling occur zoning configuration information must be updated. Mixed Zoning – Mixed zoning combines the two methods above. Using mixed zoning allows a specific port to be tied to a node WWN. This is not a typical method. WWN 50:06:04:82:E8:91:2B:9E Examples: WWN Zone 1 = 10:00:00:00:C9:20:DC:40; 50:06:04:82:E8:91:2B:9E Port Zone 1 = 21,1; 25,3 Mixed Zone 1 = 10:00:00:00:C9:20:DE:56; Port 21/1

一个理想的Zoning方法：每个Zone中包含一个 HBA，一个或多个存储端口（Storage port）。
Single HBA Zoning 一个理想的Zoning方法：每个Zone中包含一个 HBA，一个或多个存储端口（Storage port）。 Nodes can only “talk” to Storage in the same zone 一个存储端口可以是多个Zone的成员。 HBA端口（HBA port）相互之间是隔离的。简化管理。 Under single-HBA zoning, each HBA is configured with its own zone. The members of the zone consist of the HBA and one or more storage ports with the volumes that the HBA will use. Two reasons for Single HBA Zoning include: Cuts down on the reset time for any change made in the state of the fabric. Only the nodes within the same zone will be forced to log back into the fabric after a RSCN (Registered State Change Notification).

存储提供（Provisioning）: LUN Masking
指派逻辑卷（Volumes/LUN）给服务器或集群。 Zone：HBA，Storage（LUN）服务器只能访问指派给它的逻辑卷（Volumes/LUN）访问控制是由存储设备完成的，而非Fabric Makes distributed administration secure A masking database Masking管理工具 GUI Command Line Switch Device (LUN) Masking ensures that volume access to servers is controlled appropriately. This prevents unauthorized or accidental use in a distributed environment. This is typically accomplished on the storage array using a dedicated masking database. A zone set can have multiple host HBAs and a common storage port. LUN Masking prevents multiple hosts from trying to access the same volume presented on the common storage port. The following describes how LUN Masking controls access: When servers log into the switched fabric, the WWNs of their HBAs are passed to the storage fibre adapter ports that are in their respective zones. The storage system records the connection and builds a filter listing the storage devices (LUNs) available to that WWN, through the storage fibre adapter port. The HBA port then sends I/O requests directed at a particular LUN to the storage fibre adapter. Each request includes the identity of their requesting HBA (from which its WWN can be determined) and the identity of the requested storage device, with its storage fibre adapter and logical unit number (LUN). The storage array processes requests to verify that the HBA is allowed to access that LUN on the specified port. Any request for a LUN that an HBA does not have access to returns an error to the server. Array Servers

存储虚拟化（Storage Virtualization）
块级存储虚拟化 LUN的集合，形成独立于底层物理存储的虚拟存储卷。虚拟SAN FC SAN上的一个逻辑Fabric，类似于计算机网络中的覆盖网络（Overlay Network）。一组节点，无论在Fabric中的物理位置如何，都可以进行通信。

容量管理（Capacity Management）
跟踪和管理存储相关的资源已指派的端口已分配的存储（LUN）资源利用文档（Utilization profile）记录过去的资源利用率帮助预测由 SAN 管理软件提供容量管理功能，还包括资源详细目录 Capacity planning is a combination of record management, performance analysis and planning. Ongoing management issues for SAN revolve around knowing how well storage resources are being utilized and proactively adjusting configurations based on application and usage needs. The key activity in managing capacity is simply to track the assets of the SAN. Objects that should be tracked include: all SAN components the allocation of assets known utilization. For example, if the amount of storage originally allocated to a host, current usage rate and the amount of growth over a period of time are tracked, we can ensure that hosts are not wasting storage. Whenever possible, reclaim unused storage and return it to the array free pool. Do not let devices remain on host ports using valuable address space. Know the capacity of the array, what is allocated, and what is free. With this data, a utilization profile can be created. This will enables report to be created based on current allocations and consumption, and allow you to project future requests. Almost all SAN management software has the capability to capture this type of data and generate either custom or “canned” reports.

性能管理（Performance Management）
What is it? 捕获系统运行特征，监测系统运行趋势提前响应或及时响应（Proactively or Reactively）规划进一步的增长范围及功能 Host, Fabric and Storage Performance Building baselines for the environment In a networked environment, it is necessary to have an end -to-end view. Each component of the system performing either a read or write will need to be monitored and analyzed. Storage administrators need to be involved in all facets of system planning, implementation, and delivery. Databases that are not properly planned for and laid-out in an array’s backend will inevitably cause resource contention and poor performance. Performance bottlenecks may be difficult to diagnose. Common causes include: Database layout can cause disk overload Server settings impact data path utilization Shifting application loads create switch bottleneck Poor SQL code causes excess I/O

FC SAN示例什么时候需要使用SAN？ SANs提供了高带宽的、块级别的（Block level）I/O能力
能够很好地满足实时应用（real time applications）的需要 Databases: OLTP (online transaction processing) Video streaming 能够很好地满足对I/O 延时和吞吐率（latency and throughput）要求高的应用可以用于融合（ Consolidate）异构的存储环境 Physical consolidation Logical consolidation Storage Area Networks can handle large amounts of block level I/O and are suited to meet the demands of high performance applications that need access to data in real time. In several environments, these applications have to share access to storage resources and implementing them in a SAN allows efficient use of these resources. When data volatility is high, a host’s needs for capacity and performance can grow or shrink significantly in a short period of time. The SAN architecture is flexible, so existing storage can be rapidly redeployed across hosts - as needs change - with minimal disruption. SANs are also used to consolidate storage within an enterprise. Consolidation can be at a physical or logical level. Physical consolidation involves the physical relocation of resources to a centralized location. Once these resources are consolidated, one can make more efficient use of facility resources such as HVAC (heating, ventilation and air conditioning), power protection, personnel, and physical security. Physical consolidations have a drawback in that they do not offer resilience against a site failure. Logical consolidation is the process of bringing components under a unified management infrastructure and creating a shared resource pool. Since SANs can be extended to span vast distances physically, they do not strictly require that logically related entities be physically close to each other. Logical consolidation does not allow one to take full advantage of the benefits of site consolidation. But it does offer some amount of protection against site failure, especially if well planned.

融合示例：DAS面临的挑战烟囱式存储 Servers Servers Servers Storage
This example shows a typical networked environment, where the servers are utilizing DAS storage. This can be defined as “stove-piped” storage and is somewhat difficult to manage. There is no way of easily determining utilization and it is difficult to provision storage accurately. As an example, the server that hosts the black disks may be using 25 % of it’s overall capacity while the server hosting the blue disks may be at 90% capacity. In this model there is no way to effectively remedy this disparity. The only way information can be shared between platforms is over the user network -- this non-value-added bulk data transfer slows down the network and can consume up to 35% of the server processing capacity. This environment also does not scale very effectively and is costly to grow. Another issue in this model is administrative overhead. The individual server administrators are responsible for maintenance tasks, such as back-up. There is no way in this model, to guarantee consistency in the performance of such tasks. Storage 烟囱式存储

融合示例：SAN解决方案 Zoning and LUN masking Servers Servers Servers Array
Implementing a SAN resolves many of the issues encountered in the DAS configuration. Using the SAN simplifies storage administration and adds flexibility. Note: SAN storage is still a one-to-one relationship, meaning that each device is "owned" by a single host due to zoning and LUN masking. This solution also increases storage capacity utilization, since multiple servers can share the same pool of unused resources. Switch Zoning and LUN masking

连接示例：不好的存储连接 Server Server Switches Array Array Server Server Server
In this example, hosts and storage are connected to the same switch. This is a simple, efficient, and effective way to manage access. The entire fabric can be managed as a whole. Let us take an example of just one storage port. Access to the storage device must traverse the fabric to a single switch. As ports become needed on the fabric, the administrator may choose whatever port is open. Multiple hosts spread across the fabric are now contending for storage access on a remote switch. The initial design for the fabric may not have taken into account future growth such as this. This is only one example. Now imagine that there are dozens of storage ports being accessed by hundreds of hosts stretched across the fabric. Server Server

连接示例：解决方案将存储设备连接到Core Switch上 Server Server Switches Server Server
By moving storage to a central location, all nodes have the same number of hops to access storage. Traffic patterns are more obvious and deterministic. Scalability is made easy. Server Server Array

New SAN connectivity technology: IP  IP-SAN
FC SAN 面临的挑战 Infrastructure New, separate networks are required. Skill-sets As a relatively new technology, FC SAN administrative skills need to be cultivated. Cost Large investments are required for effective implementation. New SAN connectivity technology: IP  IP-SAN The traditional answer to Storage Area Networking is the implementation of FC SAN. However, with the emergence of newer SAN connectivity technology, namely IP, trends are changing. The investment required to implement FC SAN is often quite large. New infrastructure must be built and new technical skills must be developed. As a result, enterprises may find that utilizing an exiting IP infrastructure is a better option. The FC SAN challenge falls into the following categories : Infrastructure - An FC network demands FC switches, hubs and bridges along with specific GBICs and cabling. In addition, each host requires dedicated FC HBAs. Software - A variety of software tools is needed to manage all of this new equipment as well as the dedicated FC HBAs. Many of these tools do not interoperate. Human Resources - A dedicated group of FC storage and networking IT administrators is needed to manage the network. Cost – Ultimately, A good deal of time and capital must be outlayed to implement an FC SAN.

NAS vs. SAN NAS和SAN产品定位的不同：
SAN系统是高速共享存储设备子网，它适合于大规模、高性能的网络，在这种网络中快速访问是关键。SAN设备价格较高，但性能较好，并且容易解决备份的问题。它适合于中大型企业。 NAS可以很经济地解决存储容量不足的问题，但难以获得满意的性能，对于关键事务应用而言，它必须使用专用的宽带网络。而且安装调试容易，易于管理。它是中小企业的选择。

小结 SAN的定义及其特征、优点 SAN的组成 SAN的连接 SAN的管理 SAN的应用 Host Bus Adapter (HBA)
Fiber Cabling Fibre Channel Switch /Hub Storage Array Management System SAN的连接 FC SAN connectivity methods and topologies Fibre Channel devices Fibre Channel communication protocols Fibre Channel login procedures SAN的管理 SAN的应用

下课！ Q&A Discussion

Management Interface Device
EMC NAS产品 NAS device consists of: Network interface device (EMC called Data Mover) Storage interface device (EMC called Data Mover) Management interface device (EMC called a Control Station) Storage connectivity mechanism Direct connect Gateway connect Storage device Storage Interface Device Management Interface Device CLARiiON Storage Processor Navisphere Manager Symmetrix FA (Fibre Channel Adapter) Service Processor

A Data Mover is a specialized hardware platform with :
What is a Data Mover ? A Data Mover is a specialized hardware platform with : Dual Intel Processors PCI or PCI-X based High memory capacity Multi-port Network cards Fibre Channel connectivity to storage arrays No internal storage devices Operates on a highly specialized Operating System, DART (Data Access in Real Time). Each Data Mover (DM) is an independent, autonomous file server that transfers requested files to clients and writes client changes to files over an existing network infrastructure and back end storage infrastructure. There can be multiple DMs (up to 8 in the NSX frame), which are managed as a single entity by the Control Station. A DM connects to a network infrastructure via FastEthernet or GigabitEthernet and has multiple ports for this connectivity. There is no remote login capability on the DM, nor do they run any binaries (very secure). The only interface for management and configuration is the Control Station. The EMC NAS system also allows for active (primary) Data Movers to be protected by another Data Mover, of the same configuration, in the same frame as a standby. In the event of a primary failure the Control Station will facilitate a failover to the standby until the primary can be repaired. This promotes continued availability of data in the event of a failure.

What does a Data Mover Do?
Once a Data Mover has been presented storage from a storage array, the storage is divided up using an Automated Volume Management feature of DART to provide volumes for file system creation. The Data Mover supports both NFS and CIFS protocols simultaneously and is able to server out data to either protocol individually or data to both protocols at the same time. In order to share data out to differing clients simultaneously the Data Mover integrates into the security structures of both environments seamlessly Once a Data Mover has been presented storage from a storage array, the storage is divided up using an Automated Volume Management feature of DART to provide volumes for file system creation. Once these file systems have been created, mounted and exported or shared (depending on the protocol used)--via the Control Station--a client, configured with the correct protocol on a network infrastructure, can start to use the storage for data location.

Fibre I/O module GbE I/O module
Data Mover Summary NS500 Data Mover NS700 Data Mover These are the most common types of Data Mover used in EMC NAS equipment. NSX Blade Fibre I/O module GbE I/O module

What is a Control Station ?
A Control Station is a dedicated management, Intel processor based computer running EMC Linux to provide: Specialized software installation and upgrade portal Management of high availability features Fault monitoring Fault recovery Fault Reporting (CallHome) Management of Data Mover configuration and storage for the system configuration database Remote diagnosis and repair The Control Station is a dedicated management Intel processor-based computer that monitors and sends commands to the blades. The private network connects the two Control Stations (always shipped on NSX systems) to the blades through the system management switch modules. Some EMC NAS systems support two Control Stations for redundancy and continued management of the system. However if a Control Station fails without a standby system, all operational Data Movers will continue to function without an impact to their clients, only configuration and management tasks will be affected until the Control Station is repaired Like previous versions it provides software installation and upgrade services, and high-availability features such as fault monitoring, fault recovery, fault reporting (CallHome), and remote diagnosing. Two Control Stations can be connected to a public or private network for remote administration. Each Control Station has a serial port that connects to an external modem so that the Control Station can call home to EMC or a service provider if a problem should arise.

Control Station System Management – Command Line
The Control Station provides both management interfaces to NAS administrators: Command line can be accessed on the Control Station via: An ssh interface tool (e.g. PuTTy) Telnet (not enabled by default for security) Its primary function is for the scripting of common repetitive tasks that may run on a predetermined schedule to ease administrative burden It has approximately 80 UNIX command-like commands: nas_ - Generally for the configuration and management of global resources server_ - Generally for the configuration and management of Data Mover specific resources fs_ - Generally for special file system operations The most powerful interface is the Command Line Interface. Its primary function is to perform one-off tasks, or to create scripts to perform multiple repetitive tasks in one operation. The preferred access method to the Control Station for the command line is the ssh shell, as enabling Telnet increases the risk of unauthorized access and it is not enabled by default. The forms of commands within the Celerra environment are: nas_ , server_ and fs_ .

Control Station System Management – GUI Management
The other Celerra specific interface is the GUI access methodology of the Celerra Manager. This is built into the Linux operating system that runs on the Control Station and is launched through a web browser session. There are two version of this interface, Basic and Advanced, that are installed alone with the EMC Linux on to the Control Station. The options are enabled with the purchase of additional licenses to access differing levels of feature support.

Celerra NS Family – Control Station Hardware
Here is an example of Celerra NS Family Control Station hardware.

Joining the Building Blocks - Integrated
Data Mover IP Network Internal Network Switch For an Integrated configuration the system is assigned to dedicated NAS storage. This means that there are no other SAN hosts connected to the storage and therefore the whole array is dedicated solely to NAS provisioning. In the case of EMC NAS the Control Station is a separate system running an EMC flavor of Linux. This is connected to the Data Movers (client network data provider) via both an internal Ethernet and serial interface connection. The Data Mover has two internal network connections for redundancy Each Data Mover is directly, dual connected to the Storage array via Fibre Channel connections in the Integrated configuration. Once the storage array has been connected and the specialized operating system is loaded the storage array physical disks are partitioned via commands from the Control Station to create the system volumes and the data volumes. Once the specialized operating system (DART) has been loaded the Data Movers can be connected to the client network in preparation for data sharing and the Control Station for remote management. The storage array in an integrated NAS solution is an EMC CLARiiON. Control Station Serial cable connect Internal network connect Fibre Channel storage connect

Joining the building blocks - Gateway
Data Mover Fibre Channel Switch NAS NAS NAS NAS NAS NAS IP Network Internal Network Switch SAN SAN SAN SAN SAN SAN For a Gateway configuration the NAS system is assigned separately apportioned storage within the array. This means that any capacity remaining within the array can be assigned to conventional SAN hosts connected to the Fabric Switch, once the appropriate zoning and LUN masking has been performed. In the case of EMC NAS the Control Station is a separate system running an EMC flavor of Linux. This is connected to the Data Movers (client network data provider) via both an internal Ethernet and serial interface connection. The Data Mover has two internal network connections for redundancy Each Data Mover is dual connected to the Storage array via one or more (for redundancy) Fibre Channel switch(es) in the Gateway configuration. Once the storage array has been connected and the specialized operating system is loaded the storage array physical disks are partitioned via commands from the Control Station to create the system volumes and the data volumes. Once the specialized operating system (DART) has been loaded the Data Movers can be connected to the client network in preparation for data sharing and the Control Station for remote management. LUNs from the storage array can be prepared for other SAN host connectivity and then these hosts can be attached to the array once the appropriate zoning and LUN masking has been performed, to access these SAN volumes. In a gateway NAS solution, storage array can be an EMC Symmetrix or an EMC CLARiiON. Control Station SAN Host(s) Serial cable connect Internal network connect Fibre Channel storage connect

 Check Your Knowledge What are the physical elements of DAS?
Give an example of when DAS is a good solution. Describe internal DAS connectivity. Describe external DAS connectivity. What are some areas that you need to consider as part of DAS management?

 Check Your Knowledge What are the differences between a General Purpose Server and a NAS Device? What are the components of a NAS device? What protocol is used to connect to and manage physical disk storage resources in a NAS system? Give an example of a file sharing protocol. What is the difference between an integrated NAS system and a gateway NAS system?

第三章网络存储技术 Networked Storage

Similar presentations

Presentation on theme: "第三章网络存储技术 Networked Storage"— Presentation transcript:

Similar presentations

About project

Feedback

Log in

Auth with social network:

第三章 网络存储技术 Networked Storage

Similar presentations

Presentation on theme: "第三章 网络存储技术 Networked Storage"— Presentation transcript:

Similar presentations

About project

Feedback

第三章网络存储技术 Networked Storage

Presentation on theme: "第三章网络存储技术 Networked Storage"— Presentation transcript: