Changed rate estimation report can't be right

[mikeely]

For example, we've got a host holding about 160GB of stale mailboxes. The data is almost entirely static, but the report estimates deltas of over 200GB/day.

Can someone please speak to how blocks might be getting dirtied on this host? For example, does a change in atime dirty the block? What else can we look for to find out why the estimator is so very far from plausibility?

[Shestakov]

Hello Mike and welcome to the community!
Big change rate is expected for mailboxes, however 200GB/day change looks too big for 160GB VM.
There is also VM change rate history report which uses another metrics, so it`s worth comparing the results.

[mikeely]

Thank you. I'll look into the VM change rate history. In the meantime, can you speak to what metrics are being used, and if possible any strategy to avoid blocks being dirtied unnecessarily?

I don't see change rate history, only change rate estimation. What's the navigation path to it?

[mikeely]

Also I should point out that we use maildir, so the deltas should be only as big as the individual emails, not some big blob of a file that changes constantly.

[Shestakov]

Even new timestamp causes the change of the whole data block. Please read the provided link for an explanation.
VM change rate history report is in Veeam Backup & Replication Report Pack.
Thanks!

[mikeely]

Per the Extension Modules page, the Veeam Backup & Replication report pack (version 2) is installed.

How do I navigate to this report?

[Shestakov]

You need to open Veeam ONE Reporter > "Workspace" tab > All folders/Veeam Backup and Replication > VM change rate history report

[mikeely]

There is no Veeam Backup and Replication subfolder. Here is a list of what I see:
http://i.imgur.com/HVzhigP.png

[Shestakov]

I see. This means Veeam Backup Server is not probably connected to Veeam ONE.
Please add VBR Server to Veeam ONE console to enable the report pack.

[mikeely]

Ah, I see. We don't yet have that infrastructure installed, as we are using Veeam One to gather sizing data for implementation sizing purposes.

[Shestakov]

Since you already have Veeam ONE installed, why not to add Veeam Backup Server?
Without connection to VBR i.e. backup data, there is no option to use Veeam Backup reports.

[mikeely]

I've considered it, at least to test against our top talkers. We have a fairly complex environment so it's not entirely easy to just add a system in.

The main goal for right now is to determine how much storage to buy in a few weeks.

[Shestakov]

For the files size estimation you may also try a tool our system engineers use.
What are usual increment sizes of the VMs?
Out of curiosity, how do you monitor backups?

[Vitaliy S.]

My 2 cents on the topic: the report in question is based on the write rate performance counter. If data is written to the same block, it will be counted twice. Unfortunately, there is no way to estimate changed blocks without actually querying vSphere CBT driver (to do this you need to take a snapshot of the VM first). Given that for changed data estimation it would not be possible to create and delete snapshots on production VMs, that's why our engineering department has come up with this engine for estimation. While for highly transnational applications these report might return not 100% accurate values, for most of the VMs it should be close to real values.