Looking for Some Tips on Monitoring Health of Our Server

[mooseracing]

As we fill our HA setup up more, I've definitely noticed a slow down. What is everyone doing to keep an eye on stats ?

For example I can stare at Perf Mon all day on our iSCSI hosts, NIC Data, HD queues, etc, and pull a little useful data out.

We currently are running on about 2TB and 16GB's of memory. I would like to upgrade the RAM, but I don't want to take down the images. Dynamic memory allocation would be a nice feature I'm also waiting on some budget unfreeze to finish utilizing 10Gb NIC's and skip the 10Gb switch.


The reason I ask is I am not seeing queues out of the ordinary yet activity seems to hit a bottle neck once and awhile.

[Anatoly (staff)]

What are you using StarWind with? VMware? Hyper-V?

About performace - can you provide us with some numbers?

Please take a close look at our Benchmarking Guide

[mooseracing]

We are using Starwind HA for Hyper V Clusters and I can say once I set things up correctly wiring wise, I have been very happy being able to restart for maintenance and not take down any access to employees.


i did the initial benchmarks before the servers went fully live so I know what to expect throughput wise, but it's not like I can shutdown the VM's and run that test now. I was more looking for something like disk queue so I can monitor the RAID set to make sure its not being over worked, or so I can plan on upgrades.

Should I treat it just like regular storage and watch the disk queuing on the Starwind side? I know I can track it on the VM side but that doesn't help me pinpoint why it is being delayed (netowrk, disk, cpu, etc.)

[anton (staff)]

Not sure if I or anybody else here did understand what your "disk queue" actually is... Please clarify and we'll do our best to make it part of 5.8 or other upcoming release (assuming it's something reasonable and not seldom asked of course).

We are using Starwind HA for Hyper V Clusters and I can say once I set things up correctly wiring wise, I have been very happy being able to restart for maintenance and not take down any access to employees.


i did the initial benchmarks before the servers went fully live so I know what to expect throughput wise, but it's not like I can shutdown the VM's and run that test now. I was more looking for something like disk queue so I can monitor the RAID set to make sure its not being over worked, or so I can plan on upgrades.

Should I treat it just like regular storage and watch the disk queuing on the Starwind side? I know I can track it on the VM side but that doesn't help me pinpoint why it is being delayed (netowrk, disk, cpu, etc.)