#How to monitor multivolume workloads for % of system utilization?

1 messages · Page 1 of 1 (latest)

waxen igloo
#

We have a customer where they wish to monitor workloads individually to determine how much of the system resources are consumed by the workload. Customer has their own monitoring systems using Harvest with Graphite and Grafana as the repo and visualization.

Question: Is this simply a custom dashboard, combining the counters for the multiple volumes and also displaying this against the total system resource loading? Is there a way to combine all the volumes for the workload into a single conumption metric?

Tagging @wary kindle who has the account.

quartz robin
#

@waxen igloo Harvest 2.0 doesn’t use Graphite. Is the customer using Harvest 1.6? That version is no longer supported. With Harvest 2.0, we have a Workload dashboard. Please check if it helps.

image.png
wary kindle
#

Hi Rahul, Yes the customer has historically used 1.6. They understand 2.0 doesn't support Graphite - and are exploring exporters for InfluxDB and Prometheus to their in-house Graphite instance. They don't have security approval to introduce a different TSDB.

quartz robin
#

Harvest 1.6 is no longer supported. They will need to plan to move to Harvest 2.0 if version 1.6 does not have the information they are looking for.

wary kindle
#

Hi Rahul, yes they understand that.

However, the question is more how best to aggregate the performance of several volumes - to determine what percentage of the capabilities of a controller - or the percentage of the optimal point - that workload consumes under load.

The customer is trying to understand how many of their customers workloads they can fit into an HA-pair.

Is that something you can speak to? Or can point Phil and I to someone you recommend?