SQL Server On SANs
Download as PPTX, PDF0 likes1,051 views
The document discusses SQL Server management on storage area networks (SAN), including its components, performance advantages, and potential drawbacks. It emphasizes the importance of monitoring and testing configurations to prevent bottlenecks and improve performance. The content also provides practical SQLIO commands for testing SAN setups and highlights design considerations for optimal database file organization.
SQL Server On SANs
- 1. SQL Server onStorage Area Networks
- 2. Brent OzarSQL Server MVPfor Quest SoftwareManaged SQL, VM, 80tb SAN for $7b coPublished authorMade lots of expensive mistakeswww.BrentOzar.com@BrentO on Twitter
- 4. The AgendaInside the SANDBA Pros & ConsStarting From The BeginningHow to Use SQLIOMultipathing
- 7. Storage Area NetworksIndependentExpandableWell-ProtectedBlazing FastComplexExpensive
- 9. Photo Licensed with Creative Commons from http://www.flickr.com/photos/tophost/2247031208/
- 10. Photo Licensed with Creative Commons from http://www.flickr.com/photos/tophost/2246209421/
- 12. SAN ComponentsDrive enclosuresControllers (& Cache)Switch networksHost Bus AdaptersHBA DriversServers
- 15. Good ThingsClusteringTable partitioningFaster buildsEasier growthCaching speedPhotoSource http://www.flickr.com/photos/pcka/3240903696/
- 16. Not-So-Good ThingsBlack boxCachingmay not helpCompatibility listsMore driversTough to growSAN adminsmay not helpPhoto Source http://www.flickr.com/photos/stewf/95709873/
- 17. “How many gigs do you need?”
- 18. “We’ll keep an eye on it for ya.”Perfmon changesNeighbors changeSAN changesSAN monitoring software is toughPhoto Source http://www.flickr.com/photos/mrjoro/39470992/
- 19. What’s Your Bottleneck?Not Enough DrivesDrives Too SlowShared DrivesRAID LevelSAN ControllersPartition AlignmentDriversMultipathing
- 20. Things to MonitorHard drivesControllers (CPU, caching)Controller connectivity to the networkNetwork switch throughputServer connectivity to the SAN (the HBA) Server drivers
- 21. The X Factor: RedundancyHard drives – hundredsControllers – 2+Controller connectivity to network – 4+Network – separate networks, switchesServer connectivity to SAN – 2+Network fabrics – iSCSI, Fiber, WAN
- 22. SAN MonitoringStrict HCLs, SCLsCross-platformmeans JAVANo “full” pictureVery, very, veryexpensivePhoto Source: http://flickr.com/photos/earthandeden/865393933/
- 24. Start From The BeginningGet block sizes rightAlign partitionsDesign for pathingTest with SQLIOInstall SQL ServerMonitor sec/read, writeSave history
- 29. SIMPLE?!?1#%&@!?
- 30. Photo Licensed with Creative Commons from http://www.flickr.com/photos/creatista/204265480/
- 33. Active/Active GotchasOne path per LUN/volumeSending vs receivingPost-failover stickinessNetwork design
- 35. Pathing Really Matters Now
- 36. Photo Licensed with Creative Commons from http://www.flickr.com/photos/thatguyfromcchs08/2300190277/
- 37. Photo Licensed with Creative Commons from http://www.flickr.com/photos/atomdocs/3127663764/Why Can’t We Go Faster?
- 38. How to Run SQLIOsqlio -kW -t2 -s120 -dM -o1 -frandom -b64 -BH -LS Testfile.dat
- 39. Write This Down. It’s Important.sqlio -kW -t2 -s120 -dM -o1 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t2 -s120 -dM -o2 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t2 -s120 -dM -o4 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t2 -s120 -dM -o8 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t2 -s120 -dM -o16 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t2 -s120 -dM -o32 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t2 -s120 -dM -o64 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t2 -s120 -dM -o128 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t4 -s120 -dM -o1 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t4 -s120 -dM -o2 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t4 -s120 -dM -o4 -frandom -b64 -BH -LS Testfile.datsqlio -kW -t4 -s120 -dM -o8 -frandom -b64 -BH -LS Testfile.dat
- 40. We Did the Hard Work
- 41. The OutputE:\Program Files (x86)\SQLIO>sqlio -kW -t2 -s120 -dM -o1 -frandom -b64 -BH -LS Testfile.datsqlio v1.5.SGusing system counter for latency timings, -1361967296 counts per second2 threads writing for 120 secs to file M:Testfile.dat using 64KB random IOs enabling multiple I/Os per thread with 1 outstanding buffering set to use hardware disk cache (but not file cache)using current size: 24576 MB for file: M:Testfile.datinitialization doneCUMULATIVE DATA:throughput metrics:IOs/sec: 1539.50MBs/sec: 96.21latency metrics:Min_Latency(ms): 0Avg_Latency(ms): 0Max_Latency(ms): 572histogram:ms: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+%: 66 32 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
- 42. Common Bottlenecks1 gig = ~125 MB/sec2 gig = ~250 MB/sec4 gig = ~500 MB/sec10 gig = ~1250 MB/sec
- 43. Working Around ItData/logs/tempdbon different volumesSeparategroups of DBsSeparate DBinto filegroupsTempDB, backupsPhoto Source http://www.flickr.com/photos/doodlemonger/2667917978/
- 44. Wrapping It UpTest before deploymentLearn multipathingDesign DB files around SAN limitsMonitor foreverGet vendor helpPhoto Source http://www.flickr.com/photos/doodlemonger/2667917978/
- 45. Advanced TopicsTable partitioningFilegroups & filesSnapshot backupsStorage virtualizationStorage VMotionSAN administrationPhoto Source http://www.flickr.com/photos/thomasthomas/504369245/