How to create a badge in vCops : Write Latency on VM's / Aggregate

[mickaellombart]

Hi all,

Thanks for your communities and for your helpen.

I have a question concerning the vCops. We have a latency problems on some datastore.

The datastores are composed of differents Aggregates.

We have 1200 vm's and 16 Datastores. The version of vSphere is 5.1 and on vCops : 5.7.0

I will create a specific dashboard (i seek a good widget for top-analysis for example) : 

i wish find the (Top30) VM's per Aggregate last 6 Hours with metrics : High latency write en read.

We work with NetApp for the storage en i have instal the Netapp adapter on vCops.

But with differents badge i have no metrics choice for select Virtual machines / Aggr or NAS.

Could you me helpen ?

Greetz,

[mark_j]

If you're looking to analyze VMs, you should Analyze Virtual Machine : Virtual Disk : Aggregate of all instance : Total Latency for contention. Granted, it is aggregates however you'll still identify you high latency VMs. You could then do some cool stuff with interactions and RIFs to drill down in to the specific disks causing your problems.

[mickaellombart]

Hi Mark,

Sorry for late answer. I find finaly. I have now a dashboard with top write vm/datastore.

Details :

Widget : "Top-N analysis"

Last 6 hours

In first windows "Select Wich tag to show" --- >I select one host in Cluster compute Ressource

In middle windows "Ressource Kinds"  --- > "Virtual Machine"

In last windows "metrics" --- > I select "Datastrore" ; i choose one specific datastore ; and Write Latency (ms)

What to you think ?

Concerning your remarks :

"You could then do some cool stuff with interactions and RIFs to drill down in to the specific disks causing your problems."

What's is RIFS ?

Greetz Mark and thanks again

[mark_j]

Resource interaction files. https://communities.vmware.com/thread/466343 The dashboard you did works... defin a good start. Analyzing virtual disks is problematic because the aggregate and instance attribute group have the same metric names.. so if you run a heatmap in instance view, you'll always have a block that is the aggregate of all the vm disks.. this was a decision that the vcenter adapter developers made that still haunts us.. aggregate counters should have different names in instantiated groups. So yes, you can do a heatmap, but you'll have a block for each resource that is the SUM.. which kinda obfiscates the real data you're looking for and misdirects you attention.