• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

disk performance on SQL server hosted in VMware server

G

goodcooper

[H]F Junkie
Joined
Nov 4, 2005
Messages
9,771
i want to virtualize a 2k8 server with a small sql database, a hosted program we use here, and "Remote Applications" feature of 2k8 for terminal services type hosting of that app....


i was told that i'd see a lot of slow downs if i hosted this box in a virtual hard drive instead of connecting to a bare metal RAID array... mainly due to how often the SQL server is queried... a lot of little reads and writes


any truth to that?
 
It's all relative.

I run vCenter with SQL Express in a VM, no problem. I run a custom app that relies on SQL 2008 and ArcGIS Server in another VM, meaning the app, SQL server, and ArcGIS server are all installed in the same VM. No problem there either. Both VMs access the same SAN, though they not running on the same LUN. There are a number of other VMs on the same cluster / same SAN, some of which use mySQL servers, again, no issues at all.

The problem I see with sweeping generalization statements like "SQL server will slow down if run in a VM" is that they are most often simply not true. Whoever initially says that usually has no concept of what actually causes VMs (not just SQL server) to slow down.

Yes, there are circumstances where SQL server will be slower, even too slow for production, when running in a VM. Will your case be one of those? Tough to say without a lot more info, but from the sound of it my gut feeling is that you will not see any performance issues.
 
It depends on what you call a "slow down". Slower than what? Certainly, running SQL Server in a VM will be slower than not running it in a VM on the same hardware. The overhead of the VM and the limited memory shown to the SQL Server inside the VM compared to not running in the VM make this universally true.

If you must run SQL Server in a VM, you'll have to very carefully configure the VM with an eye towards tuning performance. If you can do without the VM, you won't have to worry about that overhead, can enjoy a simpler configuration, and so on.

Will the slowdown be noticeable to your application? That's something we don't have enough information to answer.
 
basically what i was asking is, is there a performance degradation to installing this into a virtual hard drive... as opposed to creating my own disk array for this application...

maybe you're answering the question and i just don't realize it... the first response was talking about how the drives were hosted on a SAN, would it matter if vmware connected directly to a volume or used a virtual hard disk placed on that volume...

is the question...
 
IMHO the performance difference between RDM and VMDK is negligible. It's far more important how many spindles that LUN is on and how many other I/O requests are made to the drives that LUN is on.
 
goodcooper said:
basically what i was asking is, is there a performance degradation to installing this into a virtual hard drive... as opposed to creating my own disk array for this application...
Click to expand...
Yes, there is. The severity of the degredation depends on your access patterns and memory requirements.
 
goodcooper said:
basically what i was asking is, is there a performance degradation to installing this into a virtual hard drive... as opposed to creating my own disk array for this application...
Click to expand...

It depends. Have you run any Perfomance Monitoring on this DB? If not you should, and over an extended period of time to get a true feel for when the DB is hit hardest, how many IPOS it needs, etc. Once you have that data, you can look at your SAN/VM infrastructure and how it's setup and see if it can support that amount/type of traffic.

With the fact that it's a small DB I'm inclined to agree with Thuleman, chances are good that you won't have much if any performance loss and things will probably run fine. To be sure however you've got to do the research. Perhaps an even easier way is since you already have a VM infrastructure (so it sounds) why don't you just build a test server with a copy of the live data and have some people try it out and see how it does? That will give you the best results possible.
 
Now I'd like to ask if all of what you told me still holds true for vbox 3.0

also, what is the skinny on running DCs in a VM? Same suggestion?
Posted via [H] Mobile Device
 
Speed of a VM is all about how you architect your environment. ESX4 has very low overhead compared to running on straight hardware. The issues I normally see are bad storage configuration. You carve out 5 disks and expect them to support 10 VMs when you would have 5 physical disks in a server running locally.

Step 1 is to analyze your app. Perfmon and see what your I/O requirements are. If a SQL server needs 1,500 IOPS then you need to know that.
 
You must log in or register to reply here.
Back
Top