Shadow Protect and Disk IOPS usage..
-
So I have a VM that by its self is slow as molasses, other VM's on the same host are running just fine.
Looking at the disk usage in performance monitor there is a single service (Shadow Protect) that is just killing the usage. Here is a screenshot.
So short of removing SP (I wouldn't mind, the boss would though) what else can I do to determine why SP is killing it?
-
How is it killing the storage? By doing less than 1MB/sec?... Queue seems fine as well.
-
Ok, did the math, seems like its reading at ~30MB/s. Again, what kind of storage do you have that's it is "killing"?.. Single hdd can cope with such speeds..
-
The IOPS for the service were and now have calmed down after the startup (but at 700 IOPS) for the sole operation.
-
@Kris_K said in Shadow Protect and Disk IOPS usage..:
Ok, did the math, seems like its reading at ~30MB/s. Again, what kind of storage do you have that's it is "killing"?.. Single hdd can cope with such speeds..
This VM is sitting on an OBR10 array, and only this VM is slow on the array. At least at start up recently from "last night" according to my boss.
-
Well, it will use all the available IOPS if you don't have any way to limit it (SIOC, etc.). If it's not limited by other resources of course.
-
And that's fine, I guess the question I'm posing is why is it using everything for just this system?
Other servers have Shadow Protect as well and they don't get hit nearly as heavily at startup.
-
I don't know enough about VM at this level - so don't crusify me if my question is stupid.
But if this is an OBR10, and no other VMs are having performance issues, and you're not limiting IOPs access to the VM somehow (maybe you are?) then why assume the storage is the root of the performance issue?
As Kris mentioned, wouldn't the host give the VM as much IOPs as it wants as long as the nothing else is asking for them?
-
@Kris_K said in Shadow Protect and Disk IOPS usage..:
Ok, did the math, seems like its reading at ~30MB/s. Again, what kind of storage do you have that's it is "killing"?.. Single hdd can cope with such speeds..
Single disks "can", but often that is their limit. And if this is a single disk system, that could hit 100% of the disk capacity pretty easily. Especially if this is not sequential. A single disk can stream over 30MB/s, but in ad hoc operations often cannot sustain it.
-
@DustinB3403 said in Shadow Protect and Disk IOPS usage..:
@Kris_K said in Shadow Protect and Disk IOPS usage..:
Ok, did the math, seems like its reading at ~30MB/s. Again, what kind of storage do you have that's it is "killing"?.. Single hdd can cope with such speeds..
This VM is sitting on an OBR10 array, and only this VM is slow on the array. At least at start up recently from "last night" according to my boss.
If the disk was being saturated, seems like everything on the array would notice.
-
@Dashrender said in Shadow Protect and Disk IOPS usage..:
I don't know enough about VM at this level - so don't crusify me if my question is stupid.
But if this is an OBR10, and no other VMs are having performance issues, and you're not limiting IOPs access to the VM somehow (maybe you are?) then why assume the storage is the root of the performance issue?
As Kris mentioned, wouldn't the host give the VM as much IOPs as it wants as long as the nothing else is asking for them?
Yes, it will use as many IOPS as it can that are available. So that alone isn't indicating a problem.
-
So the question that I have is, why should a single VM hitting 600-700 IOPS really be slow while running this?
Just make no sense when everything else on the host is running fine. Nothing I was able to capture showed anything else even coming close to a high usage.
-
How are you measuring the slowness? What does it "feel" like?
-
@DustinB3403 said in Shadow Protect and Disk IOPS usage..:
So the question that I have is, why should a single VM hitting 600-700 IOPS really be slow while running this?
Just make no sense when everything else on the host is running fine. Nothing I was able to capture showed anything else even coming close to a high usage.
Maybe the problem isn't in hardware, but in a process that's running.
Now I don't understand why a process wouldn't have the processors pegged out (and perhaps a single core is pegged out - i.e. single threaded process) to allow the process to be completed as fast as possible.
-
@scottalanmiller said in Shadow Protect and Disk IOPS usage..:
How are you measuring the slowness? What does it "feel" like?
As was reported to me, an extremely long log in period. (It feels like listening to nails on a chalk board)
-
@DustinB3403 said in Shadow Protect and Disk IOPS usage..:
@scottalanmiller said in Shadow Protect and Disk IOPS usage..:
How are you measuring the slowness? What does it "feel" like?
As was reported to me, an extremely long log in period. (It feels like listening to nails on a chalk board)
Logging into the app or the server (VM)?
-
@Dashrender said in Shadow Protect and Disk IOPS usage..:
@DustinB3403 said in Shadow Protect and Disk IOPS usage..:
@scottalanmiller said in Shadow Protect and Disk IOPS usage..:
How are you measuring the slowness? What does it "feel" like?
As was reported to me, an extremely long log in period. (It feels like listening to nails on a chalk board)
Logging into the app or the server (VM)?
Logging into the VM using the console, or RDP.
-
What's the CPU usage like in the VM?
-
Nominal CPU usage.
-
Could it be waiting on something from the network?