EMC FAST Cache with SQL Server 2012 - with video

Solution homepage: https://community.emc.com/docs/DOC-14614

 

The testing of this solution validated the ability of the VNX5700 storage array to support SQL Server 2012 in instances running OLTP-like workloads that generated over 50,000 IOPS.

 

Note    Benchmark results are highly dependent on workload, specific application requirements, and system design and implementation. Relative system performance will vary as a result of these and other factors. Therefore, this workload should not be used as a substitute for a specific customer application benchmark when critical capacity planning and/or product evaluation decisions are contemplated.

 

All performance data contained in this report was obtained in a rigorously controlled environment. Results obtained in other operating environments may vary.

 

EMC used a number of scenarios to test the solution. These included:

  • Baseline testing on an 40 x SAS-only storage pool
  • Performance testing on a storage pool with FAST Cache enabled
  • Comparing and profiling reseeding times for AlwaysOn Availability Groups for the following availability modes:
    • Synchronous-commit modeAutomatic failover
    • Synchronous-commit modePlanned manual failover
    • Asynchronous-commitForced manual failover

 

Throughput in IOPS (transfers/sec)

Throughput was measured using the Microsoft Performance Monitor (perfmon) counter: LogicalDisk – Avg. Disk Transfer/sec.

FASTCacheIOPs.png

Figure 1 Avg. Disk Transfer/sec (IOPS) for both primary and secondary replicas

 

During baseline testing with 40 SAS disks in the storage pool transactional I/O throughput on the primary replica produced approximately 11,500 IOPS and the secondary replica produced 1,400 IOPS.

 

After 30 minutes with FAST Cache enabled on the storage pool we saw an immediate effect on performance. I/O throughput increased to over 19,000 IOPS on the primary replica and to 2,300 on the secondary. After just two hours of FAST Cache running we saw throughput increase to over 50,000 IOPS on the primary replica.

Throughput in transactions per sec (TPS)

Throughput was also measured using the Microsoft Performance Monitor (perfmon) counter: Databases – Disk Transactions/sec.

 

  FASTCacheTPS.png

Figure 2 Disk transactions per sec (TPS) for both primary and secondary replicas

 

During baseline testing with 40 SAS disks in the storage pool, transactional activity, shown as SQL Transactions/sec or (TPS), was over 4,900 TPS with the secondary replica at 300 TPS.

After 30 minutes with FAST Cache enabled on the storage pool, SQL Transactions/sec on the primary replica increased to approximately 12,000 TPS and to 900 TPS on the secondary replica. After just two hours of FAST Cache running we saw transactional performance increase to over 24,000 TPS.

Testing was conducted with distance emulation between production and DR sites initially set to 80 km distance during FAST Cache warm-up, with replication set to synchronous-commit mode. Distance was then stretched to 800 km and then 4,000 km, both with replication set to asynchronous-commit mode.

Latency

Latency was monitored through testing; Avg. Disk/sec reads and writes were monitored, at the following distances:

  • 80 km synchronous-commit mode
  • 800 km asynchronous-commit mode
  • 4,000 km asynchronous-commit mode

 

After EMC FAST Cache had warmed up, it allowed average latency to be maintained at 3 ms or less for reads and 1 ms for writes on the primary replica for all three distance options.