Spikes are occuring unregular. I cannot find any relation to jobs running on this or other servers. Here's a screenshot of todays metrics: http://sdrv.ms/18p5TZR
. Because this is 1 day period values are a little "foggy" - you do not see spikes that are only half of others. I counted the errors of last 24 hours: There are approx. 25 and no recognisable pattern.
[Update]: There have been some occurences last minutes with maximum values of 127131031 ms , that's 35 hours. That could not be any real latency. At the same time I had very low DAVG/cms values esxtop.
Anyway, thanks for your help, I will try to get any VMware support - or simply ignore this issue, because there is no constraint in "real life"