Presentation is loading. Please wait.

Presentation is loading. Please wait.

Physical Storage Components

Similar presentations


Presentation on theme: "Physical Storage Components"— Presentation transcript:

1 Physical Storage Components
Introduction This lesson discusses a number of component technologies relevant to the storage arena. These discussions cover disk storage subsystems, the components of a disk drive, disk driver interfaces, kinds of access and access time, and disk performance metrics. The lesson also examines array controllers, and tape storage. Importance Understanding the basic physical storage components is important when considering the overall performance of a storage system and the role of Cisco equipment in Storage Area Networks (SANs). © 2003, Cisco Systems, Inc. All rights reserved. 1

2 Lesson Objective Upon completion of this lesson, you will explain the following physical components of a storage system: disks, array controllers and tape storage devices. Performance Objective Upon completion of this lesson, you will explain the following physical components of a storage system: disks, array controllers and tape storage devices. Enabling Objectives Identify the components of a disk storage subsystem. Identify the basic components of a disk drive and explain their function Explain the most common disk drive interfaces Explain how seek time and rotational latency equal access time Explain the difference between random and sequential access Explain the significant disk performance metrics Explain the role array controllers play in a storage subsystem. Explain array controller characteristics Explain array controller architecture, including high and low-end examples Explain the key performance metrics of array controllers Explain the importance of tape storage, despite the limitations of tape and difficulties associated with tape formats Explain what a tape library is, including references to common sizes and configurations Explain the basics of linear and helical scan recording Explain tape storage advantages Explain tape storage disadvantages

3 Outline Disk storage subsystems Components of a disk drive
Disk drive interfaces Access time Random vs. sequential access Approximate rated numbers for disk performance What is an array controller? Array controller characteristics Array controller architecture Key array controller performance metrics The importance of tape storage Tape libraries Tape recording schemes Tape storage advantages Tape storage disadvantages Summary Prerequisites The qualifications to attend the program.

4 Disk Storage Subsystem
Host Interface Power Supply Controller Cache Internal I/O Channel Disk Storage Subsystem Objective Identify the components of a disk storage subsystem. Introduction This section introduces basic storage subsystems. Facts In its simplest form, a disk storage subsystem is a chassis that contains a power supply, fan, host interface, disk array controller, disk drives and a high speed internal I/O bus. The host interface attaches servers to storage using media such as Fibre Channel (FC), SCSI, Enterprise System Connection (ESCON), and Fibre connectivity (FICON) The disk drives are typically attached internally using SCSI or FC high-end storage subsystems include redundant power supplies, array controllers, caches, and I/O channels (dual ported disk drives)

5 Components of a Disk Drive
Platters are made of aluminum alloy or glass substrate Disk drives contain electronics used to perform a number of functions Platters are accessed for read and write operations using two read/write heads Components of a Disk Drive Objective Identify the basic components of a disk drive and explain their function. Introduction This section provides an overview of the physical components of a disk drive. Facts Disk drives store data by magnetically recording on rotating disks (“platters”) using a read-write head. The platters are made of aluminum alloy or glass substrate. The media layer is a very thin coating of magnetic material, which is where the actual data is stored, and is typically only a few micro-inches in thickness. The surface of each platter is normally covered with an extra-thin, protective, lubricating layer, on top of the magnetic media layer itself. As a disk rotates, a circular track occurs under the read-write head. Disk drives contain many separate tracks available for storing data. Each track is divided into segments called sectors or blocks. These are the basic unit of data storage on a disk drive.Most disk drives contain more than one platter. When the vertically aligned tracks on the platters are taken together, they make up a cylinder. Disks rotate at 5400, 7200, 10,000 or 15,000 rpm. Platters are accessed for read and write operations using two read/write heads, mounted on the top and bottom of each platter. The heads are mounted onto arms that allow them to be moved from the outer tracks of the hard drive to the inner tracks. The arms are controlled using a device called an actuator that positions the arms to the appropriate track on the disk. The read/write heads do not touch the platter when the platter is spinning; instead they float on an extremely thin cushion of air (10 millionths of an inch). Continued…

6 Disk Drive Interfaces Disk drives may use any of a number of different interfaces: ATA/EIDE Serial ATA SCSI Bus Fibre Channel Disk Drive Interfaces Objective Explain the most common disk drive interfaces Introduction This section provides a brief overview the disk interfaces used in various storage environments. Facts The connection between a system or controller and disk drive may use any one of a number of different interfaces. The interface chosen depends on the environment. The IDE interface is the most common disk interface, followed by SCSI and Fibre Channel. Definitions ATA(Advanced Technology Attachment )/EIDE (Enhanced Integrated Drive Electronics) (ATA/EIDE ): Low-cost interface for devices used in personal computers. Serial ATA: A new, serial version of the ATA/EIDE interface by IBM. SCSI Bus: Principal interface traditionally used in high-performance workstations and servers. Fibre Channel: A high-end interface also used for high-performance workstations and servers. Provides more connectivity and distance than the SCSI bus.

7 Access Time Access Time = seek time + rotational latency Access Time
Objective Explain how seek time and rotational latency equal access time. Introduction This section introduces the two components that comprise disk access time: seek time and rotational latency. Facts Each platter has its own read-write head used to access data on that platter. Because the read-write head can only be in one place at a time, it is necessary to move the head to the correct track before reading or writing data. This process is referred to as “seeking” and the time required to complete the movement is the seek time. The faster the arm positioning, the faster the seek time. Once the head is positioned at the correct track, it is necessary to wait for the desired sector or block to rotate to a position under the head (remember, the disk is rotating and the sectors or blocks pass under the head as the disk rotates). This time is referred to as the rotational latency and, on average, is equal to one half of the disk’s rotation time. Therefore, the faster the disk rotates, the smaller the rotational latency. Taken together, the seek time and rotational latency comprise the access time—how long it takes to access data on the disk. Depending upon the drive model, this time typically ranges from 5 to 10 milliseconds. This makes disk drives ideal for data that must be accessed rapidly. Continued …

8 Random vs. Sequential Access
Disk drives provide the ability to access data in two ways: A random manner A sequential manner Random vs. Sequential Access Objective Explain the difference between random and sequential access. Introduction This section discusses the difference between random and sequential access. Facts Disk drives allow data to be accessed in a random manner. The disk drive is given an instruction to read or write data beginning at a designated location, and the head seeks location and begins reading or writing. This type of access is called random access. Tape drives, by contrast, require that reading or writing start at the beginning of the tape and progress sequentially along the tape. This type of access is called sequential access. Some types of data applications require random access. Most databases require random access as they retrieve requested information. Think how inefficient it would be to sequentially scan through every record in order trying to find a specific piece of information. When random access is used with disk drives, each access requires a seek (with the associated overhead). Because of the overhead associated with seeking, performance, as measured in megabytes per second, is usually lower with random access when compared to sequential access. Other types of data applications use sequential access. Video stored on disk drives is normally arranged so that the data can be accessed in a sequential manner—this minimizes the amount of time spent seeking and maximizes the amount of time spent transferring data. Consequently, performance is higher when using sequential access.

9 Approximate rated numbers for disk performance*
Description Rated IOPS Rated throughput Rated latency 7200 rpm 80-100/sec 10 MBps sequential 7.5-9 MBps random 8–10 ms rpm /sec 18-20 MBps sequential 15-17 MBps random rpm 150/sec 26-28 MBps sequential 25 MBps random 6–8 ms Disk Performance Objective Explain the significant disk performance metrics. Introduction This section covers approximate rated numbers for disk. These are important for an SE to know, as they relate to potential storage bottlenecks in a client's storage system. Facts This chart covers basic disk performance metrics. It is important to understand basic disk performance metrics, because they might indicate the source of a performance bottleneck that can affect overall storage system throughput. Generally speaking, faster is better. IOPS refers to the maximum I/O operations per second (IOPS). IOPS characterize the maximum number of small block read/write operations that a disk or storage system can sustain over a period of time. Rated throughput refers to the data transfer rate, measure in megabytes per second. Rated latency is measured in milliseconds. * Numbers may vary plus or minus ten percent, depending on the disk vendor. Source: Ravindra Neelakant, Andiamo Systems

10 What is an Array Controller?
The control logic in a storage subsystem that performs control and management functions Can be implemented in software, firmware or hardware May be integrated within an intelligent host bus adapter or external storage subsystem What is an Array Controller? Objective Explain the role array controllers play in a storage subsystem. Introduction This section gives a definition of an array controller, and indicates what an array controller does. Definition A storage array controller consists of the control logic in a storage subsystem that performs command transformation and routing, data integrity (Redundant Array of Independent Disks [RAID], mirroring, striping, or other), high-level error recovery, and performance optimization for multiple storage devices. Facts The Storage Network Industry Association ( defines a storage array as a collection of disks or tapes from one or more commonly accessible storage subsystems, combined with a body of control software that provides common control and management facilities. The control software presents the arrays of disks or tapes it controls to its operating environment as one or more virtual disks or tapes. Control software may execute in a disk controller or intelligent host bus adapter, or in a host computer. When it executes in a disk controller or adapter, control software is often referred to as firmware. Continued …

11 Array Controller Characteristics
Improved performance Reduced server and network overhead Array Controller Characteristics Objective Explain array controller characteristics. Introduction This section provides an overview of array controller characteristics. Facts Storage array controllers provide many of the high-level features that contribute to availability, reliability and scalability within modern storage environments. Modern array controllers typically utilize RISC-based processors (i960, PPC, etc.), high-speed internal buses, and local cache chips, all of which are optimized for disk I/O. SCSI-based array controllers can achieve bandwidth up to 320 MBps per channel (SCSI Ultra320). FC architectures can attain up to 200 MBps bandwidth. SCSI over IP (iSCSI) solutions will soon be available and will offer even higher bandwidths. (see Module 1, Evolution of Storage Solutions, Lesson 1, Storage Protocols for a discussion of SCSI, FC and SCSI over IP). Array controllers can contribute additional performance improvements to servers and the SAN. Server performance is increased by off-loading processor and memory otherwise used for disk I/O. Network traffic is reduced when data is mirrored as the data only needs to pass through the SAN a single time. This data is cached by the controller and then independently written to each member of the mirrored volume.

12 Array Controller Characteristics (cont.)
High availability and fault tolerance Redundancy (RAID 1, 4, 5, etc.) Local and remote data replication Fault prevention & data recovery Virtual disk volume support Enhanced security Array Controller Characteristics (cont.) Facts Many high availability and fault tolerant features can be implemented at the storage array controller level, including: RAID level 1 (disk mirroring), RAID level 4 (data striping at a block level across several drives, with parity stored on one drive), and RAID level 5 (data and parity striping at a block level across several drives). The ability to recover the lost data from a local disk. On high-end disk subsystems (for example, the EMC Symmetix Remote Data Facility) data replication can occur across geographically distributed networks and remote arrays. Remapping of faulty disk sectors. Virtual disk volume support can be implemented in the array controller, allowing multiple physical disk partitions to transparently appear as a single logical volume to one or more servers. The controller converts SCSI or FC requests for a logical block into one or more requests for physical disks. Advanced controllers offer enhanced security, for example, the ability to associate specific logical unit numbers (LUNs) or FC World Wide Names (WWNs) with Access Control Lists (ACLs).

13 Array Controller Architecture
Server Storage Subsystem Application File System Array Controller Array Controller Architecture Objective Explain array controller architecture, including high and low-end examples. Introduction This section provides an overview of array controller architecture, introducing both high and low-end examples. Facts The diagram describes a generic path from application to disk array. For example, when a file is saved in an application such as Microsoft Notepad, the data must traverse the file system, device driver and hardware interface to reach a physical disk drive within the storage array. In general, the following steps must occur: An application makes a request to read or write data to or from disk The file system converts these requests to read/write data from/to disk data blocks The driver converts the request into SCSI commands, which are then transported to the host bus adaptor The host bus adapter encapsulates the commands and data into the transport medium for that particular HBA The array controller converts the SCSI requests made for virtual disks into requests for physical disk blocks Device Driver HBA Interface

14 Array Controller Architecture (cont.)
Typical parallel SCSI disk array controller architecture Fibre Channel Arbitrated Loop also commonly used for attached devices Array Controller Architecture (cont.) Facts Historically, disk array controllers have been store and-forward devices. Host computers send I/O commands and data to the controller, which receives them in their entirety and relays them to their destinations. The diagram illustrates a typical parallel SCSI disk array controller architecture. The controller’s internal bus connects the host and disk interfaces with the controller’s cache (cache sizes can range from 4MB or less, up to 1GB or more). Connection options between the host computer and the array controller’s host interface include SCSI, FC and Ethernet. The two might be attached directly or through a fabric switch. Consider, for example, what occurs when a host’s read command is first received by the controller’s host interface. The controller’s processor schedules the internal operations required to execute the host’s command. For example, the processor converts each host read command addressed to a disk into an equivalent SCSI command. The SCSI command is sent to the SCSI interface connected to the target disk. The disk interface and the disk itself execute the command and deliver data to the controller’s cache. When the data is available in cache, the processor schedules transmission to the host, completing the host’s SCSI task. More complex tasks, such as reading data from a RAID array, require more extensive processing. For example, if a host’s read requires data from two disks, the processor must schedule two disk read operations and ensure that all data has been delivered before it signals completion of the host’s original request. A key aspect of this architecture is intermediate storage of host commands and data in controller cache. Receiving and reprocessing host commands is how a controller adds its value, particularly in RAID arrays.

15 Array Controller Architecture (cont.)
Direct attached storage Limited scalability Drive redundancy only Single physical location Virtual disk volumes RAID 0, 1, 5, etc. Array Controller Architecture (cont.) Facts In the early days of RAID, storage arrays were built using internal or direct attached storage (DAS), as illustrated in the diagram. In this configuration, the array controller resides on the server, most frequently integrated directly into the host bus adapter. This configuration is still popular for low-end storage arrays. For example, intelligent host bus adapters that plug into a PCI slot are available with Parallel SCSI or FC Arbitrated Loop (FC-AL) interfaces from a number of vendors including Adaptec, Mylex, Compaq, HP and many others. While there are many benefits to this configuration (virtual disk volumes, mirroring, striping, etc.) there are also limitations. For example, the number of attached drives is limited (up to 15 drives per SCSI channel or 127 drives per FC-AL), and the storage distance from the server is limited (up to 25 meters for SCSI or 10 km for FC without extenders). Although there is no server redundancy displayed in the diagram, it would be possible to use a multi-initiator SCSI to attach two servers to the same array. In addition, the fault tolerance can be increased in the server by using dual host bus adapters and dual-ported disk drives. Low-end Configuration Example

16 Array Controller Architecture (cont.)
Array controllers integrated in storage subsystems (EMC, HP/Compaq, HDS, NetApps, IBM, etc.) External storage Expandable Fully redundant Remote replication Virtual disk volumes Security Integrated Management FC Switch Array Controller Architecture (cont.) Facts At the higher end of the spectrum are externally attached disk storage subsystems from vendors such as EMC, HP/Compaq, Hitachi Data Systems, Network Appliances, IBM, and others. These high-end storage subsystems provide many (two or more) multi-ported array controllers and dual-ported SCSI or FC disk drives. They deliver full redundancy, virtually unlimited scalability, security (for example, through LUN mapping), support for remote data replication (for example, EMC’s SRDF), integrated management facilities, and improved performance. FC Switch Higher-end Configuration Example

17 Key Array Controller Performance Metrics
Protocol/Media type: Maximum throughput, cable length, and number of devices: Parallel SCSI 320 MBps per channel, 25 m, 15 devices Fibre Channel 200 MBps (2 Gbps), approx 10 km, 16M nodes (2^24) iSCSI 100 MBps (1 Gbps), unlimited, 4B nodes (2^32) Key Array Controller Performance Metrics Objective Explain the key performance metrics of array controllers. Introduction This section provides an overview of key array controller performance metrics. Facts The table reviews the current performance capabilities of various media types used to connect storage array controllers and devices. These are theoretical maximums. For example, not all addresses are available in the FC and IP address spaces. Throughput represents the maximum speed at which data can be delivered to an array controller. Maximum distances do not include the use of cable extenders and repeaters. In the not too distant future, Ethernet speeds should increase even further.

18 The Importance of Tape Storage
Tape storage is based on recording data on magnetic tape Tape is predominantly used for backup and restore, and archiving The Importance of Tape Storage Objective Explain the importance of tape storage despite the limitations of tape and difficulties associated with tape formats. Introduction This section introduces the learner to the continued importance of tape storage, despite the inherent disadvantages of tape storage. Facts Most backup and recovery systems use a tape device of some kind that provides the primary storage function of backups. Tape backup, once widely criticized for being too slow, is still important in storage environments. High-end tape backup units that use DLT have reasonable performance. Generally speaking, it provides low-cost data storage with a wide range of performance and data storage capacities. A tape drive requires a longer time to access data than does a disk drive. For one thing, tape is a removable medium and may require human intervention to locate and mount the tape containing the data (this could take minutes). To reduce the mount time, many installations use tape loaders or automated tape libraries that use a robotics mechanism to mount the tape. Once the tape is mounted, the tape drive must position the tape to the correct location before reading or writing data. Depending on the speed of the device, this may take several seconds. Continued … Robotic Tape Backup

19 Tape Libraries Tape libraries can hold thousands of tapes and have multiple tape drives. Tape Libraries Objective Explain what a tape library is, including references to common sizes and configurations. Introduction This section provides a basic overview of tape libraries. Facts Tape libraries are tape automation products that are larger than and hold more tapes than autoloaders. They often have more than one tape drive, although that is not a requirement. Tape libraries can be quite large and hold thousands of tapes. Removable tape magazines are common in these systems. There are a number of possible configurations and geometries used in library systems.

20 Tape Recording Schemes
Linear/Serpentine Recording Helical Scan Recording Linear and Helical-Scan Recording Objective Explain the basics of linear and helical-scan recording. Introduction This section describes the tape recording schemes known as Linear and Helical-Scan Recording. Facts Overview There are two basic recording schemes in modern use: linear recording and helical scan recording. Magnetic recording is performed by moving an electro-magnetic head past a magnetic media (such as tape) at high speed. Generally, the higher the head-to-tape speed ratio the better. The higher speed provides an improved signal-to-noise ratio. The improved signal-to-noise ratio results in improved data reliability, increased data rates, and higher bit densities. There are differences between linear and helical-scan tape drives. The most notable difference is the method used to write data to the tape. In linear recording, data is recorded to the tape by rapidly moving the tape across a stationary head. By comparison, helical-scan devices record data by keeping the tape nearly stationary while moving the heads. The different methods result in overall system design and performance differences. Continued …

21 Tape Storage Advantages
Performance of high-end units is adequate Provides low-cost bulk storage Archived data can be stored off-site Ability to restore multiple versions Tape Storage Advantages Objective Explain tape storage advantages. Introduction This section discusses tape storage advantages. Despite the inherent disadvantages of tape storage, it is still popular. Facts Tape storage is popular for the following reasons: High-end tape backup units that use DLT have reasonable performance. It provides low-cost data storage with a wide range of performance and data storage capacities. Because of the access time, tape is used primarily where low-cost bulk storage is needed but not fast access time. Examples of tape applications include data backup, archival, and infrequently accessed data. Tape does have one distinct advantage when compared to disk storage—it is removable. Following a backup operation, the tape can be removed from the site for safe keeping where it is both physically and operationally secure. Redundancy provided by RAID does not protect against computer room disasters or operational failures. Mirrored data is of no use if an application or the operating system corrupts both copies. The ability to restore multiple versions is important if, for example, systems have been infected by a virus.

22 Tape Storage Disadvantages
Performance Data access time Tape Storage Disadvantages Objective Explain tape storage disadvantages Introduction This section discusses tape storage disadvantages. Despite the continued use of tape storage, there are still some significant disadvantages. Facts Disadvantages of tape storage are: Performance is limited. Even high-end devices cannot match the performance of disk arrays. A tape drive requires a longer time to access data than does a disk drive. For one thing, tape is a removable medium and may require human intervention to locate and mount the tape containing the data (this could take minutes). To reduce the mount time, many installations use tape loaders or automated tape libraries that use a robotics mechanism to mount the tape. Once the tape is mounted, the tape drive must position the tape to the correct location before reading or writing data. Depending on the speed of the device, this may take several seconds.

23 Random vs. Sequential Access Practice
Answer the following questions. Be prepared to share your answers with the class. 1. Explain the difference between random and sequential access. 2. For which types of applications is sequential access appropriate? Random vs. Sequential Access

24 Disk Drive Components Practice
Match the components on the left with the correct item on the right read-write head covered with a very thin coating of magnetic material tracks where data is stored disk electronics moves the read/write heads into position records data onto a disk platter microprocessor and firmware data buffer actuator I/O interface sectors or blocks where data is recorded Components of a Disk Drive

25 Disk Drive Interfaces Practice
Give a brief explanation of the following interfaces: ATA/EIDE Serial ATA SCSI Bus Fibre Channel Practice: Disk Drive Interface

26 Tape Storage Practice Answer the following questions:
What are the primary uses of tape storage? What is a tape library? What are the two tape recording schemes? What are the differences between them? List advantages and disadvantages of tape storage. Practice: Tape Storage

27 Lesson Practice Answer the following questions. Be prepared to share your answers with the class. 1. What is? a disk storage subsystem an array controller a tape storage device 2. What is the role of each of the above in a storage system? Practice: Physical Storage Components

28 Summary This lesson covered these key points:
The basic physical storage components are: disk storage subsystems array controllers and, in some cases, tape storage devices Physical Storage Components Review In this lesson, you learned about a number of component technologies relevant to the storage arena. The topics included disk storage subsystems, the components of a disk drive, disk drive interfaces, access time, and disk performance. The lesson examined array controllers, including characteristics, architecture, and performance as well as tape storage basics.

29


Download ppt "Physical Storage Components"

Similar presentations


Ads by Google