Ultra high disk latency within guest OS, nothing bad shown on datastore

[breakaway9000]

Hello,

I'm having some serious issues with the performance of my virtual machines. I've been on a wild goose chase recently running all over the place but I think I've foudn the problem.

My datastores inside of ESXi are showing disk read/write latency times of <20ms at most, but my VMs (windows server 2008/2012) are showing in-OS disk response times (as shown in resource manager) of between 100 and 700 ms. Suffice it to say, everything is grinding to a halt.

Does anyone know why the VMs may show such high disk queue lengths, but vmWare doesn't show this, and how I can start getting to the bottom of this?

[zXi_Gamer]

Can you confirm if swapping is not the root cause of this, by checking on esxtop.

[MarVista]

Hello,

Have you tried to Defrag Virtual Hard Drive?

Hope will help.

Yours,
Mar Vista

[breakaway9000]

Hi,

I'm not very good with esxtop -- can you tell me what I'm looking for here?

Cheers

[zXi_Gamer]

In your esx box command line,

1. type esxtop and press Enter

2. Now press M and press "Shift + V"

3. Look for the values under SWCUR and SWTGT. Ideally, these should be 0

As usual, this Interpreting esxtop Statistics is an excellent read about esxtop

[Paltelkalpesh]

Use Veeam Monitor to consolidate disk usage and latency for all hosts in a single chart. The free version of the product allows you to consolidate up to 24 hours,

[DavidSilva77]

Hello buddy, this kind of issue in the %90 of the cases is causes by problems in the SAN, if you SAN uses Fiber Channel you must examine the zonning of the switches, examine the ports and storage, generally this problems is not causes by hosts ESXi or vCenter