Test performance/utilization of a backup & replication environment

[jmc]

Hello everyone,

we have a quite extensive backup environment with many different repositories, proxies & agents.

i would like to have an overview of the utilization and over- or underprovisioning of the different components to see if e.g. a proxy has too few processors or memory, if the number of tasks on a repository is possibly too high.

in other words, everything you need to know to monitor and optimize a backup environment.

are there any existing ideas or ways?

thanks
jeff

[david.domask]

Hi jeff,

I think the Backup Infrastructure Report from VeeamOne is what you're looking for:

https://helpcenter.veeam.com/docs/one/r ... ml?ver=120

If it's missing some elements for you, there is also the community supported veeam-healthcheck tool: https://github.com/VeeamHub/veeam-healthcheck

It should work out of the box, but please note that the veeam-healthcheck does not receive support, and issues will ned to be addressed on the github page.

[jmc]

hi david,

i know veeam one and the tool is not bad. unfortunately it is primarily focused on backups and jobs and not on what i am looking for.

i would like an overview of whether all proxies have enough or too much cpu, memory, network connection, etc. for the veeam environment. or whether the repositories would run better with 6 tasks instead of 10.

the best thing to do is to monitor whether a job can run behind another job or in parallel with one. yes, after a data change it can happen that a job has to wait for another one because the resources are not free, a vm is blocked or the number of snapshots of a datastore is exceeded and the reg-key has to be adjusted.

ok, the whole thing is really asking a lot , but asking if something like this already exists is not reprehensible.

btw:
something like this may of course cost something.

at the moment i'm making do with a specific dashboard in aria operations. but i can only record everything from the vmware perspective.

but thanks for the tip. the tool is really useful.

thanks
jeff

[david.domask]

Hey Jeff,

Got it, I think I get the idea, though a few points I just want to clarify:

When you mention too much CPU/Memory/NetworkConnection, you want to find machines that are over-provisioned for their workload? (Network one I'm not quite getting in full; I think you're meaning a situation where the proxies/repos are underperforming compared to previous runs, but just want to make sure I got this correct)

As I'm understanding it, you're asking for a "tuning" tool of sorts that will understand/suggest if you need more resources, fewer tasks, etc; am I correct?

[jmc]

hello david,

yes, you are right. i just want to monitor the design to see if there is anything that could be improved. for example, i can see the enpass for a job (source, destination, ...) for example, another evaluation would be useful that says that the line is the problem because too many simultaneous tasks are working. or the proxy lacks the ram to work optimally.

this is also about both directions. both over- and undersized. it makes no sense if a proxy has too much cpu or ram allocated and it doesn't need them. or that a backup could be finished faster if i increase the number of tasks on a repository a little.

i'm thinking of monitoring the individual components in the design. recording the data of the backup server, the proxies and repositories. check the bandwidth and latency of the components among each other. the whole thing is then best viewed by ai on the jobs and time. so a statement could then be made as to whether job a should not better run before job b, because at time a, for example, too many snapshots are available and this reduces the overall performance of the source.

vmware has the aria operations for this, something like this would be just right for veeam. there are countless evaluations of the jobs and the design.

what is the best thing to do at the moment if a job takes too long or has to wait too long for resources? you install another proxy. but this is not the best solution in all cases.

maybe the development team will get bored and take care of the problem. it would certainly be a good extension to veeam one.

But there is also a side effect. A special customer of ours has issued environmental compliance in order to save ressources. the best way to do this is through power consumption. Therefore, all systems are currently being checked to see if there is anything that can be saved. If there is a program that tells me that another proxy is not necessary if you just do the following, then that would be great too.
(just as a result: during the check, 2 of 17 esxi hosts were already switched to standby.)

jeff