How much impact does hard drive cache have on IO performance
Are there any statistics or performance tests available that can tell me if there is a difference is 16MB vs 32MB Cache?
If it matters, I am looking at SATAII 7200RPM Drives.
Server Fault is a question and answer site for system and network administrators. It only takes a minute to sign up.Sign up to join this community
Hasn't this question been asked before?
Basically if you're doing lots of reads from a 17-32Mb data set and have no local cache then it'll make a difference - for everything else you might see a 1-2% overall increase in performance but you're also doubling the chance that in the event of a power loss your disk integrity is lost as the cache isn't battery-backed.
I would pay attention more to performance reviews of hard drives of the same size and rotational speeds (7200, 10000). I have not seen the same series of hard drive manufactored in 2 different cache versions. There is normaly other performance improvements beyond a cache increase in higher end models that make them faster. It is not the cache alone that could make that big of difference.
I would not go out of your way to get a 32mb cache version unless you wanted the higher performance models, which generaly come with higher cache memory already.
Caching algorithms, latency, areal density, etc., will all impact your performance--in some cases, they might affect performance even more than the actual cache size. You need to look at benchmarks for various hard drives in the application you're considering.
You can find some good benchmarks here: http://www.storagereview.com