Hi,
Any one knows how vRealize calculate IOPs which are shown in report? Is it for average for month or how is it?
Thanks in advance.
It depends on the report as they will use different views. Edit the report and see what view it is using and then edit the view
Under data it will tell you what it is using (Transformation)
Under Time settings it will tell you the time frame it is using
So this view is showing the Max, Min, and average over the last 30 days
Hi,
Not able to find option you specified
Edit the report template not the generated report. That will tell you the view it is using
Where to go next?
The middle column lists all the views that make up the report.
scroll down to the view that has the IOPs table and get the name of it.
then on the left column find the name of the view and select it
click the pencil
You are now editing the view. Any changes you make to that view will change any report or dashboard that is using it.
Thanks
It's for 7 days. If report is showing average IOPs for VM A is 250 for 30 days & I need to size storage how much IOPs I need to consider?
250 IOPs for 30 days ÷ 30 Days ÷ 24 hours ÷ 60 min ÷ 60 Sec?
It all depends on who you ask
I would not look at each VM separately and i would not use the average over 30 days either. I would be looking at the max for all VMs over a 30 day period (longer if possible) At the end of the day your storage needs to provide enough IOPs for all VMs running on it. Your average for all VMs could be 40K Iops but on one day of the month it goes up to 60k due to an AV scan or something you dont want all your VMs to suffer because the storage cant provide that amount of iops. But in some cases this might be acceptable if the scan is at the weekend and no one is using the VMs.
It also depends on the SAN and how the data is held and how the SAN provides IOPs. If the SAN uses some auto tiering and virtual LUNs the you dont need to worry about providing enough IOPs per datastore because the SAN will do that for you. But if the SAN has a one to one relationship between spindles and datastores then you would need to know the IOPs per VM so you dont over load a LUN