I love data. I like seeing numbers and graphs. I like to see if something is having an impact.
I like when new capabilities provides us with the means to monitor because this data gives me reassurance that the feature has an impact instead of me simply believing it does.
Let’s look at XenServer 7.1 and Provisioning Services Accelerator. I was able to show that
But I didn’t have any details beyond what I was able to gather from a stopwatch.
That was until I starting poking around XenCenter. I was thrilled to see a set of metrics specific to PVS Accelerator
I ran a few tests to see what I could gather from this data. The results were what I was hoping to see.
First, I looked at network traffic between the PVS server and XenServer host.
On the flip side, I took a look at the amount of network traffic saved with PVS accelerator.
- Without PVS Accelerator, my maximum throughput during boot was 30,000,000 bytes
- With PVS accelerator, my maximum throughput during boot was close to 60,000,000 bytes.
I said that utilizing PVS Accelerator resulted in a decrease in boot times by 50-60%. The VM must read the same amount of data, but because we have the data cached locally in RAM, we are doubling the throughput, resulting in a 50% reduction in bootup times.
The same process is happening in logon times, which is why PVS Accelerator also decreases logon times.
TIP: If you want to capture XenServer monitoring data in a .CSV file, use “rrd2csv -n -s 5 > FILENAME” from XenServer console. (-n uses label name instead of GUID, -s is sample period in seconds)