We create the Disk usage alarm in the virtual server center, but some times an alarm report that the disk usage is over 100%.
Is it a bug of Virtual Serve Center?
How to explain it?
is this measured on a "normal" VM, or is this during some kind of benchmark ?
A read-only disk read benchmark will read blocks from the virtual disk that have never been written to before. The VMkernel optimizes this by returning zeroes from memory, rather than zeroes from disk. This transfer of zeroes from memory happens faster than the physical disk channel can sustain, which makes it plausible that it would be measured as >100% bandwidth.