The Basics of Reliable Distributed Storage Networks 姓 名 : 冼炳基 學 號 :

Outline  Introduction & Why use DSN ??  What is SAN(Storage Area Network) ??  What is NAS(Network-Attached Storage) ??  Distributed SAN Application  Distributed NAS Application  SAN or NAS ??  Future directions : IP storage  Summary

Storage networks Introduction  Increase storage efficiency and data availability  Permit user quickly and efficiently perform information management function Backup and recovery Data mirroring Disaster recovery Data migration  Efficient of stored data greater than traditional “islands of storage”  Manage access data to increase performance, survivability and control costs

Why use DSN ?? (1)  Geographical and distances Ensure Data survivability over wide geographical areas Provide data synchronization over large distances  New regulatory requirements Sarbanes-Oxley Act Health Insurance Portability and Accountability Act Securities and Exchange Commission  Serial optical-fiber-based storage protocols Escon (Enterprise System Connection) Fiber Channel

Why use DSN ?? (2)  Bus-based protocols limited to a few meters  Native-mode Escon and Fiber Channel transmission more than 10 kilometers  Repeaters or link extenders as far as 100 kilometers  Storage data travel over MANs and WANs using transport protocols ATM (Asynchronous Transfer Mode) Sonet (Synchronous Optical Network) WDM (Wavelength-Division Multiplexing) IP (Internet Protocol)

What is SAN (Storage Area Network)  One common storage network  Computing and storage nodes via fabric of network switches  Switched fabric support multiple simultaneous full-bandwidth connections  A common backup server connect to any storage device for scheduled or manual backup  Employ a gateway provide MAN of WAN interworking and protocol translation

SAN implementation

What is NAS (Network-Attached Storage)  Consists of a file manager attached to a LAN  NAS difference with SAN NAS serves file-structured data to clients SAN serves block-structured data to application servers  Some NAS implementation NAS head to serves files to its clients Attaches to a back-end SAN to manage block-structured data  Two issues for grographic range of NAS Distance limitations inherent in LAN protocols Use of a distributed SAN to provide backend storage for a NAS

NAS architectures

Distributed SAN Application  Storage integration  Remote backup and restore  Disk mirroring  Data migration  Business continuity and disaster recovery  Remote operation of peripheral devices  Mainframe and open-systems connectivity

Storage integration  Using DSN to share disks and intergrate storage across a wide grographic area  Useful in amount of stored data is large Database of genetic information Multimedia or video servers Multiple servers update a common database  User don’t know where the data’s acutal physical location  Must limit data access to one user at any time Employ a locking mechanism

Shared-storage configuration

Remote backup and restore  Non-real-time backup and restore from a remote location  Traditionally remote backup defect Time-consuming Disruptive process Dedicated personnel and equipment  Back up automatically across the MAN of WAN to tape or disk  Restore data from the backup copy across the MAN of WAN

Remote backup and restore

Disk mirroring  Near-real-time mirror data to multiple disk  Mirroring vs. backup data availability cost  Synchronous vs. Asynchronous mirroring  Local system log change immediately then periodically logged change to remote system  Forms of disk mirroring Processor-centric Storage-centric Split-mirror

Processor vs. storage -centric

Split-mirror  Use a third mirror disk Increase reliability Minimize the possibility of data loss  Normal operation vs. copying data operation  Optimum performance and minimal disruption  The better solution for perform copy operation Read-only can continue normally Frozen image  Two basic ways to resynchronize third mirror Overwrite all data to third disk from the other two Use a transaction log to update changed blocks

Split-mirror architecture

Data migration  Efficiently move large volumes of data  Previously an awkward process Need to copy volumes of data to type Physically transprot to new location Disrupting day-to-day processing  Storage networks provide an elegant solution Mirroring or backup can create a new copy at new location New location copy can intend of old location copy at a specified time

Business continuity and disaster recovery  DSN are important for this require  Loses data Using standard backup and restore mirroring  Loses processing capability Processing continue by secondary site  Either site becomes inoperative Other site can continue full processing

Remote operation of peripheral devices  Use storage protocols to remotely orperate peripheral devices Printers Check sorters  Perform remote-peripheral operation SAN Mainframe storage architectures

Mainframe and open-systems connectivity  Application for connectivity Integration of e-commerce and mainframe-based application Data warehousing Backup and recovery  Moving data among mainframe and open systems LAN FTP  Transfer the data directly between storage subsystems is faster Gateway or bridge translates between Escon or Ficon Ficon storage protocol used in mainframe channel subsystem and SCSI Fiber Channel used in open systems storage network

Mainframe and open-systems connectivity architecture

Distributed NAS Application Distrubute NAS application in two way  Using switched gigabit Ethernet Enable NAS file managers to be 5 kilometers Production LAN without needing repeaters  For NAS head configuration Distribute the back-end SAN across multiple sites Using any of distribute SAN applications

SAN or NAS ??  NAS A short-term tactical solution Use an existing Ethernet LAN Performance for smaller installations of less than 5 Tbytes)  SAN For large installations(greater than 5 Tbytes Where management and reliability Multisite data sharing and replication Long-term strategic solution  SAN and NAS are beginning to converge A NAS head with a SAN back end identical to SAN NAS and SAN as two different views of the same data

Future directions : IP storage  Using internet and TCP/IP to transport block storage data Ubiquity and availability of internet technologies Cost and performance advantages of internet technologies  New application over internet iSCSI (Small Computer Systems Interface Protocol over the Internet) FCIP (Fibre Channel over TCP/IP) iFCP (Internet Fibre Channel Protocol)  Think of IP storage as a compatible technology

Summary  Distrubuted storage applications improve the efficiency  Remote backup and mirroring improve reliability  Business continuity and disaster recovery enable enterprises to recover quickly and transparently  Storage protocols and gateway enable data transfer between mainframe and open-systems  NAS provide shared file access for clients using LAN  Integrate with SAN to provide truly distributed metwork capabilities