I've got two WD5000AAKS's; 500GB SATA 3gb/s drives with 32MB caches, and two SG Barracudas 500GB SATA 3gb/s drives with 16mb cache. The question I have is: does the cache size between elements in a RAID array affect the overall performance of the array in any meaningful way? I know that platter rotational speeds, interface speeds, and even size mismatches (losing size) can have a huge bearing on arrays, but is an array going to suffer, or even not function if I mix caches? I was planning on doing RAID 5 with all four with the onboard controller on an ASUS P5N-D motherboard. I suspect at the worst case, it will simply do what it does in every other situation like this and treat all of them as 16MB cache; the lowest common value, but I am not certain if it will even matter.
If anyone has any expertise or experience with a similar setup, I'd appreciate the input.
Really there will be basically no difference between those drives in RAID from a cache perspective. On another note, I wouldn't use 4x drives in RAID5, I'd go for 3 drives or 5 drives. It has to do with how the RAID/Parity is distributed and has a pretty big effect on performance depending on which controller you're using (when it comes to onboard controllers, which are software, it's a big deal--which happens to be your case). Also matters how you create the array and the settings you choose. If you're using Vista, this is easier, but if using XP for example, it's more involved.
Anyhow, the cache's won't make a difference in this case.