DCAWD.com

I have an application question for those who may have run into this realm already.
We are using Virtual Workstation to run server VMs. The servers (6G virtual ram with 2GB host file separation size, ~64GB total) are running IIS to serve web service calls to a SQL database that uses replication services to external clients. Lots of small data writes and IO functions for replication checking...
We are finding that we run into performance issues that are not RAM dependent when running more than 2 VMs from the same 7200 RPM 64mb cache HDD. We can run two additional VMs on another physical disk, so it seems as if the IO read requests are too much for the spindle/cache seek rate of the drive.
Looking at a Seagate hybrid 8gbSSD/1TBHDD also with a 64mb cache. Not so much for the data transfer rate, but the ability for regularly used data blacks to be cached for faster IO functions.
Question that I have is will the SSD hybrid portion be able to be populated with frequently accessed data blocks to enable an additional VM to be run, or is that population done only with the host OS indexed files, leaving VM operation out of the mix entirely?

For an application like that, you may want to look in to something like this:

http://www.fusionio.com/products/

They have products for accelerating VM, memory and cache applications, may be a better fit than just a hybrid drive.

The SSD portion you don't control, it's there as a secondary cache. You probably would see an improvement, but with SSD's coming down in price, why even bother. Mirror two SSD's and call it a day. You'll get twice the read performance and write performance will go up from your present setup. The problem you're running in to is IOPS, which we see a lot in VDI deployments.

Remember that VMWare VM's use a large VMDK file, so it's pretty hard to cache since it's one big file to the host OS (there are exceptions where SAN's/NAS's are VM aware), so going to a dedicated SSD in your case is the best bet.