I usually see this if there's been a problem with the ESX host or the the VC server.

Are the missing gaps just for one ESX Host, one VM Guest or for all hosts/guests at the same time?

If it's for just one ESX host, then it's probably a host problem, either it diconnected, there was a network outage, check the service console nic speed/full duplex settings are correct. Possibly try restarting the vmware-ccagent.

Just a few initial thoughts.

John

Has the service console got enough memory allocated to it?

try top/esxtop to see if anything is happening there.

I get the same across all 13 host servers and 150 VMs, but only for the smallest timeframe - (1 hour) which is not configured by default. The default minimum timeframe is over 1 day and I added this other shorter timeframe to monitor hosts/VMs on.

I believe that larger timeframes average the smaller ones to get their average - ie if monitoring every minute for an hour, this is averaged out for the hourly entry within a day which is in turn averaged out for the daily over the week and so on...

I don't tend to worry about it too much as the way in which VirtualCenter seems to manage its database, the data won't be there in the next hour anyway and there are no issues looking at other timeframes (the bigger picture of what's been happening over the last week/month).

I don't know if any alerts at the host level in VC (ie CPU %) are affected by this, but I have those configured to alert for any dangerously high usage and I have to date been alerted when the appropriate conditions are met.

There do seem to be a few bugs/inconsistencies with VC (latest version that I'm using), so here's holding out for a significant update!...

I've not heard anything definite about any future releases (content/dates/etc..), however i have heard the rumour *unofficially* that the way in which monitoring works may change to the structure more reminiscent to other enterprise management applications, whereby instead of VC (the application) going to an ESX host and checking its condition, it would be the other way round, so that the host notifies VC when it has something to notify on.

I have also segmented each of my ESX host's console NIC on a separate VLAN from a core Cisco switch and put VC on this VLAN, yet I still get these breakages in performance data....

Roll on a future release of ESX/VC!...

My experience has been the same (almost exactly) as mtrouardriolle. I took it a step further though. My VC host was running as a VM and I moved it to a physical box to try to resolve the issue. It seemed to improve things but it still has the gaps in the data, just not as big or as much (sometimes i would be missing an entire hour before wheras now I will only miss up to 20 mins at a time). Truthfully however, the problem seems to be related to upgrading my ESX host version as well as my VC. I cannot pin the issue down however...

We're not running VC under ESX but as phi. machine.

Gaps still remaining even if network throughput is actually reduced for the fact many users are on vacation... continues to sound me so strange.

Seems there is something to do if that issues compares: to clear perf. statistic and force VC to reload ...

I don't actually know if it works but I've read something about that in another forum tread...

Oh, so that Access Database I'm still having to compress every week or 2 may be the cause?