By Reid Smith-Vaniz
Is the performance of your EMC array not meeting your expectations? Here are nine proven remedies for that. What will work for you depends on your workload, configuration, network and more. So look through the whole list first, find the remedies that apply to your situation, and take steps to get the performance you were expecting from EMC.
- Add disk drives. If you don’t have enough disk drives in your system to support the workload, that will adversely impact performance. Adding more drives allows the system to read and write the data more quickly.
- Run RAID 10. There are different RAID configurations: 1.0, 5, 50, and others. There’s a trade-off between performance and capacity. RAID 10 is the best-performing configuration. If you’re running a demanding workload like Oracle on RAID 5, you might consider shifting it to RAID 10.
- Switch from large to small drives. When you’re running a high-performance application, you’re better off adding more drives of a smaller size, like 300GB, than using very large drives, like 2 TB drives, 900GB SAS drives, or 600GB Fibre Channel drives. With smaller drives, the drive actuator doesn’t need to travel as far to fetch data, so performance improves.
- Switch from SATA to SAS, FC, or solid state. For high-performance workloads, like VMware, switch your storage from SATA drives to SAS, Fibre Channel or solid state—all three are faster than SATA. You may have started out using your SATA drives to store data for workloads that don’t demand high performance, but over time, you may have begun using those SATA drives to store data for demanding applications. If that’s the case, you’ll have a performance issue.
- Uncrowd the storage array. You may have too many applications on a single array, and performance suffers because there aren’t enough drives or processing capability to support the workload. The physical capacity may be sufficient, but the performance won’t be. Put the right applications and the right number of applications on each storage array
- Upgrade the controller. You may have the proper number of drives and the right RAID configuration, but the controller may be a bottleneck. Replacing the controller with a newer one may eliminate the performance issue. (But note that this is not an easy remedy; others listed here are easier.)
- Peel out the workload. Place a couple of applications on another storage array or upgrade the array entirely. If you have too many applications on an array, you can consider expanding its capacity. But if the system is pushing up against its capacity limits, you need to remove applications.
- Check the SAN fabric. You may have bought a new storage array that’s capable of running at 8GBs per second, but it’s connected via 4 GBs per second SAN fabric. Take a look at the performance diagnostic tool within your SAN management console to see if you have latency in the SAN fabric. If so, the fabric may be your performance bottleneck.
- Replace your 4Gbps HBAs. Make sure your HBAs aren’t the bottleneck. You may not have HBAs in all your servers, but you should consider them for servers with applications that are I/O intensive, such as Oracle, SQL or VMware servers.
Good luck with tuning and improving the performance of your EMC array. If you’d like advice tailored to your exact situation, just give our storage experts a call. 877.227.0828.