Getting the most performance impact from a flash purchase involves more than just buying fast hardware or efficient caching software. Both of these components must work together to maximize their use of processing power that’s available.
The hardware certainly must generate as many IOPS as possible to the flash layer, which means an ideal design that includes low latency and high parallelism. But if the hit ratio to the cache is low then the application is going to wait for data to be fetched from the disk layer, wasting those available IOPS. This means CPU cycles must be ‘spent’ on the software process that decides which data is cache worthy, in order to eliminate cache misses.
Minimizing resource overhead of the caching software while still maximizing its effectiveness to squeeze the most performance out of the hardware is a balancing act that’s not easy to pull off. Systems that integrate the caching software and flash hardware are best able to walk this tightrope. OCZ does this, but also takes the next step by optimizing its operation for the SQL environment with the ZD-XL SQL Accelerator.
This PCIe card and software creates a turnkey solution that DBAs can install easily and get maximum performance from their SQL applications. In this video George Crump and Allon Cohen, OCZ’s VP of Software and Solutions, discuss the ZD-XL and Direct Pass Caching technology and how they’re optimized for the SQL environment.
