You may have noticed over the weekend that the wizards behind Infosight have been busy with a rather cool facelift to the Infosight User Interface (if you haven't had a look, go now - i'll wait for you.....)
One of the coolest things about this update is the new Per-VM Monitoring feature! This is found under the Manage->Virtual Environment tab.
There are a few requirements for the VM Monitoring will work, which are:
- You must be using VMware vSphere 5.x or above on your hosts.
- You must have NimbleOS 220.127.116.11 or above on your Nimble arrays.
- You must have the Nimble vCenter plugin provisioned.
We fully support use of either the Windows vCenter implementation or the vCenter Server Appliance. Infosight can also monitor and display multiple vCenter instances (really cool if you have multiple site deployments of vSphere/Nimble!).
If you meet all three of these requirements, you are free to enable the Per-VM Monitoring feature yourself, of course with no charge. Here's how to do it:
Within Infosight, browse to the Administration-> Virtual Environment tab:
You will then see a list of your arrays, your NimbleOS versions, and whether or not you have the vCenter plugin provisioned. You can also see if "streaming data" is enabled or disabled; this is where the Nimble array is streaming the vCenter statistics back to Infosight. In my example below, both are disabled.
That's it, we're done! Within 24 hours the Nimble array will start to stream the collected vCenter stats to Infosight, and data will start to be processed for output on Infosight. It may take 24-48-72 hours for the data to start displaying, so please don't worry if it doesn't show up right away.
One of the Infosight wizards took the time out from being awesome to record this highly useful demo of the new Infosight UI & the Per-VM Monitoring feature, which is available on Youtube here:
How Does It Work?
The Nimble array uses the vCenter plugin to pull various statistics about each ESXi host, and correlates them on the Nimble array. Every hour or so those correlated stats are then sent to Infosight via the heartbeat mechanism for processing; meaning there are no additional host-side agents/tools or licenses to administer for this feature to work.
What Are The Use Cases?
We're now using the vCenter collection API to get the following:
- VM latency
- Host CPU, Memory and latency
- VMDK IOPS, MBPS and latency
- Datastore IOPS, MBPS latency
Meaning we can then use our Infosight data analytics and case creation service to start performance troubleshooting:
- Identify overloaded hosts (CPU, memory, and balloon activity)
- Visually explore the latent VMs by datastore
- Identifying VM noisy neighbours
- Find Nimble volumes for VMware datastores
Note: Any approved Nimble partners are not permitted to view any customer's VM data.
As usual, please post questions, thoughts or feedback below - the wizards will be watching!