Network Communication and Technology Mod
- 1. Remote Replication Section 3 : Business Continuity
- 2. Chapter Objective • Explain remote replication technologies – Synchronous and asynchronous • Discuss host and array based remote replication – Functionality – Differences – Selecting the appropriate technology • Discuss network options for remote replication
- 3. What is Remote Replication? • Replica is created at remote site – Addresses risk associated with regionally driven outages – Could be a few miles away or half way around the globe • Modes of remote replication (based on RPO requirement) – Synchronous Replication – Asynchronous Replication Source site Remote site
- 4. Synchronous Replication • A write must be committed to the source and remote replica before it is acknowledged to the host • Ensures source and remote replica have identical data at all times – Write ordering is maintained • Replica receives writes in exactly the same order as the source • Synchronous replication provides the lowest RPO and RTO – Goal is zero RPO – RTO is as small as the time it takes to start application on the target site 1 3 4 2 Data Write Data Acknowledgement Host Target Source
- 5. Synchronous Replication: Bandwidth Requirement • Response Time Extension – Application response time will be extended • Data must be transmitted to target site before write can be acknowledged • Time to transmit will depend on distance and bandwidth • Bandwidth – To minimize impact on response time, sufficient bandwidth must be provided at all times • Rarely deployed beyond 200 km Time Writes MB/s Required bandwidth Typical workload Max
- 6. Asynchronous Replication • Write is committed to the source and immediately acknowledged to the host • Data is buffered at the source and transmitted to the remote site later – Some vendors maintain write ordering – Other vendors do not maintain write ordering, but ensure that the replica will always be a consistent re- startable image • Finite RPO – Replica will be behind the source by a finite amount – Typically configurable 1 4 2 3 Data Write Data Acknowledgement Host Target Source
- 7. Asynchronous Replication: Bandwidth Requirement Response time unaffected Bandwidth Need average bandwidth Buffers Need sufficient buffers at the source to hold data before transmission to remote site Can be deployed over long distances Average Time Writes MB/s Required bandwidth Typical workload
- 8. Remote Replication Technologies • Host based – Logical Volume Manager (LVM) based • Support both synchronous and asynchronous mode – Log Shipping • Storage Array based – Support both synchronous and asynchronous mode – Disk Buffered - Consistent PITs • Combination of Local and Remote Replication
- 9. LVM-based • Duplicate Volume Groups at source and target sites – All writes to the source Volume Group are replicated to the target Volume Group by the LVM – Can be synchronous or asynchronous mode • In the event of a network failure – Writes are queued in the log file and sent to target when the issue is resolved – Size of the log file determines length of outage that can be withstood • Upon failure at source site, production can be transferred to target site IP
- 10. LVM Based – Advantages and Limitation Advantages The source and target sites do not need to use the same storage arrays and RAID protection configuration Response time issue can be eliminated with asynchronous mode (with the drawback of extended RPO) Specialized hardware / software is not required because LVM is available in the OS running on the host Limitations Extended network outages require large log files If log files fill up before outage is resolved, full synchronization is required Replication process adds overhead on host CPUs Remote host must be continuously up and available
- 11. Host Based Log Shipping Offered by most database vendors Transactions to the source database are stored in logs Log files are transmitted to the remote host when the DBMS switch log files Happens at pre-configured time intervals or when a log file is full The remote host receives the logs and apply them to remote database Advantages: Minimal overhead Low bandwidth requirement Standby database is consistent up to the last applied log
- 12. Host Based Log Shipping Original Logs Standby Logs IP
- 13. Source Array Storage Array Based Remote Replication • Replication performed by the array operating environment – Host CPU resources can be devoted to production operations instead of replication operations – Arrays communicate with each other via dedicated channels • ESCON, Fibre Channel or Gigabit Ethernet • Replicas are on different arrays – Primarily used for DR purposes – Can also be used for other business operations Target Array Distance Source Replica DR Server Production Server IP/FC Network
- 14. Array Based – Synchronous Replication Network links Write is received by the source array from host/server Write is transmitted by source array to the target array Target array sends acknowledgement to the source array Source array signals write complete to host/server Source Target
- 15. Array Based – Asynchronous Replication • No impact on response time • Extended distances between arrays • Lower bandwidth as compared to Synchronous Write is received by the source array from host/server Write is transmitted by source array to the target array Source array signals write complete to host/server Target array sends acknowledgement to the source array Network links Source Target
- 16. Asynchronous Replication: Ensuring Consistency Maintain write ordering Some vendors attach a time stamp and sequence number with each write, then send the writes to remote array These writes are applied to the remote devices in exact order based on the time stamp and sequence numbers Dependent write consistency Some vendors buffer the writes in the cache of the source array for a period of time (between 5 and 30 seconds) At the end of this time current buffer is closed in a consistent manner and the buffer is switched, new writes are received in the new buffer Closed buffer is then transmitted to the remote array Remote replica will contain a consistent, re-startable image on the application
- 17. Array Based – Disk Buffered Replication • Local and Remote replication technologies can be combined to create consistent PIT copies of data on target arrays • RPO usually in the order of hours • Lower Bandwidth requirements • Extended distance solution Source Storage Array Target Storage Array Local Replica Remote Replica Local Replica Source Host Source Data
- 18. Remote Replicas – Tracking Changes • Remote replicas can be used for BC operations – Typically remote replication operations will be suspended when the remote replicas are used for BC operations • During business operations changes will/could happen to both the source and remote replicas – Most remote replication technologies have the ability to track changes made to the source and remote replicas to allow for incremental re-synchronization – Resuming remote replication operations will require re- synchronization between the source and replica
- 19. Array Based – Which Technology? • Synchronous – Is a must if zero RPO is required – Need sufficient bandwidth at all times – Rarely above 125 miles • Asynchronous – Extended distance solutions with minimal RPO (order of minutes) – No Response time elongation – Generally requires lower Bandwidth than synchronous – Must design with adequate cache/buffer capacity • Disk buffered – Extended distance solution with RPO in the order of hours – Require lower bandwidth than synchronous or asynchronous
- 20. Three Site Replication • Eliminates disadvantages of two site replication – Single site disaster leads to a window when there is no DR protection • Data replicated to two remote sites • Implemented in two ways – Three Site Cascade/Multi-hop – Three Site Triangle/Multi-target
- 21. Three Site Replication – Cascade/Multi-hop • Synchronous + Disk Buffered • Synchronous + Asynchronous Bunker Site Remote Site Local Replica Remote Replica Local Replica Source Data Synchronous Remote Replica Disk Buffered Source Site Bunker Site Remote Site Local Replica Remote Replica Local Replica Source Data Synchronous Remote Replica Asynchronous Source Site
- 22. Three Site Replication – Triangle/Multi-target Asynch with Differential Resynch SOURCE REMOTE BUNKER Sync Async SAN SAN S A N
- 23. SAN-based Remote Replication • Replicate from one storage array to any other storage array over the SAN/WAN – Implement tiered storage – Data migration – Remote vaulting • Heterogeneous arrays support • No impact to servers or the LAN Hitachi EMC Symmetrix EMC CLARiiON HP IBM SAN/WAN
- 24. SAN Based Replication : Terminologies • Control Array: Array responsible for the replication operations – Control Device: Device on controlling array to/from which data is being replicated • Remote Array: Array to/from which data is being replicated – Remote Device: Device on remote array to/from which data is being replicated • Operation – Push: Data is pushed from control array to remote array – Pull: Data is pulled to the control array from remote array Control Array Remote Array PUSH PULL Control Device Remote Device C C SAN/WAN
- 25. Network Infrastructure • A dedicated or a shared network must be in place for remote replication – Use ESCON (Enterprise System Connectivity) or FC (Fibre Channel) for shorter distance – For extended distances, an optical or IP network must be used – Example of optical network: DWDM and SONET • Protocol converters may require to connect ESCON or FC adapters from the arrays to these networks – Native GigE adapters allows array to be connected directly to IP Networks
- 26. Dense Wavelength Division Multiplexing (DWDM) • DWDM is a technology that puts data from different sources together on an optical fiber with each signal carried on its own separate light wavelength • Up to 32 protected and 64 unprotected separate wavelengths of data can be multiplexed into a light stream transmitted on a single optical fiber ESCON Fibre Channel Gigabit Ethernet Optical Channels Optical Electrical Optical Lambda λ
- 27. Synchronous Optical Network (SONET) • SONET is Time Division Multiplexing (TDM) technology • Traffic from multiple sources of different speeds are organized into a frame and sent out onto SONET ring as an optical signal • Synchronous Digital Hierarchy (SDH) similar to SONET but is the European standard • SONET/SDH offers the ability to service multiple locations, and provides reliability/availability, automatic protection switching and restoration SDH STM-1 STM-16 SONET OC3 OC48
- 28. Chapter Summary Key points covered in this chapter: • Modes of remote replication – Synchronous and asynchronous mode • Host based remote replication – LVM based and log shipping • Array based remote replication – Synchronous, asynchronous and disk buffered – Three site replication – SAN based remote replication • Network options for remote replication
Editor's Notes
- #1: Chapter 14 – Remote Replication