I want to setup a small ESXi server, which has to be silent, therefore I will use a SSD drive, however I am not sure which one I shall use and which parameters of a SSD are important for usage with ESXi?
I am thinking on one of these (price rougly the same, capacity enough for me)
- OCZ Onyx 32GB
- Intel SSD 320 Series 40GB
- OCZ Vertex 30GB
- Kingston SSDNow V-Series 30GB
For example the Onyx is faster in writing than the Tntel SSD, but the Intel SSD performs better on 4K, but I have to admit I have no clue whether 4K is important for ESXi! The Kignston seems to have an agressive cleanup / trim like behaviour built-in maybe that's useful because ESXi supports no trim?
A silent ESXi server, well there's a first time for everything I guess - are you sure this is for a professional/production system? I'd be really interested in what you're doing here.
Anyway it seems a bit odd to be a) using an SSD as your main datastore at all and b) worrying about storage performance when you'll be using a passively cooled CPU (presumably). Unless you know SSDs are the right solution for you and that your full end to end system will benefit from the extra performance these higher-end SSDs then I'd go for pretty much any of them but with an emphasis on choosing the one with the most over-commit memory.
The answer entirely depends on your workload and your requirements for reliability.
SSDs do "wear out" as they are written to. All of your picked drive models are MLC drives primarily targeted for the laptop market where writes do not happen very often for large amounts of data
If you do not expect extensive I/O with high contention from your guests, you might want to look into SLC-based drives instead - they will cost more and typically show lower performance, but will be more reliable as SLCs allow about 10x the number of write cycles compared to MLCs.
BTW: the datasheet numbers might seem impressive, but the drives often do not deliver on the promise. We've had Transcend SLC drives stating "up to 90 MB/s sequential write performance" in the datasheet but yielding actual 12-14 MB/s in the lab which was "within the expected range" according to Transcend tech support. So the basic rule here would be to test before deployment as you might get surprising results with your particular setups.