Just took over a large Veeam environment - Some Questions

[steggles]

Hi All,

I just took over a large VMware/Veeam environment and I want to get some recommendations on a few jobs that are required.

There is a requirement to backup all the file servers with 365 daily restore points. The previous engineer setup a daily Forever Forward Incremental Primary Job, with 365 restore points. From my understanding, Forever Forward Incremental backups are considered bad practice for jobs with this many restore points. How should I reconfigure this job to ensure I have 365 daily restore points, but ensure that I am meeting Veeam best practice? Do I split the Job into one 30day primary job with a backup copy job for 365 dailys? How often should I run a full backup?

The other questions relate to Windows Application Aware processing / VMware Tools quiescence. The current Microsoft Services Manager does not want to provide an account with Domain Admin rights in order for Application Aware processing of Windows Machines (but has provided accounts for SQL Server Application Aware Processing). To that end, would my second best choice would be VMware Tools quiescence for all other Windows Machines?

WRT to VMware Tools quiescence, our Unix Team Manager does not want to have this turned on for the Linux servers as they believe it will impact performance. So I am only getting Crash Consistent backups for all Linux servers which seems somewhat dangerous. Is there any reason why VMware Tools quiescence should not be turned on?

Cheers.

[Vitaliy S.]

Hi,

There is a requirement to backup all the file servers with 365 daily restore points. The previous engineer setup a daily Forever Forward Incremental Primary Job, with 365 restore points. From my understanding, Forever Forward Incremental backups are considered bad practice for jobs with this many restore points. How should I reconfigure this job to ensure I have 365 daily restore points, but ensure that I am meeting Veeam best practice? Do I split the Job into one 30day primary job with a backup copy job for 365 dailys? How often should I run a full backup?

It's not considered bad practice, but it can take some time when you decide to restore data from the oldest restore point. It's really up to you to decide if you're fine or not with the performance you're seeing. Here is an existing topic regarding the question of the full backup best practice> The need for Active Full backups

The other questions relate to Windows Application Aware processing / VMware Tools quiescence. The current Microsoft Services Manager does not want to provide an account with Domain Admin rights in order for Application Aware processing of Windows Machines (but has provided accounts for SQL Server Application Aware Processing). To that end, would my second best choice would be VMware Tools quiescence for all other Windows Machines?

If servers you're backing up have highly transactional applications, then using application-aware processing is highly recommended. If you have no choice, then you can use VMware Tools quiescence option.

WRT to VMware Tools quiescence, our Unix Team Manager does not want to have this turned on for the Linux servers as they believe it will impact performance. So I am only getting Crash Consistent backups for all Linux servers which seems somewhat dangerous. Is there any reason why VMware Tools quiescence should not be turned on?

What kind of performance impact your Unix Team Manager is referring to? Do you know any details?

Using VMware Tools quiescence with pre-freeze and post-thaw scripts is recommended to have a consistent backup of applications running on these servers. If your Unix team is already backing up critical data with native, built-in to application scripts, then you can skip the VMware Tools quiescence option. Please review this thread for further reading.

Thanks!

[steggles]

Hi Vitaliy S, thanks for the reply.

They may be somewhat out of date, but threads here (like this one) tend to indicate that primary jobs should not be used for such long retention. But these threads tend to be quite old, if the current rule of thumb is that these types of jobs are ok, I will keep the job as is.

The thread you linked regarding Linux Server backups was useful. Is there a gallery of pre-freeze and post-thaw scripts for different scenarios? (I am not a unix expert by any means, if I can approach the Unix team with premade scripts it may help them be less resistant to the using VMware Tools quiescence)

Regards,
Thomas Higgins

[Vitaliy S.]

1. Yes, these are old threads, that's why I've referenced Anton's post where he addresses this question.
2. There is no gallery/library of scripts, but application teams should be aware of the correct way to back up their data and, usually, script samples are provided by the application vendor.

[skrause]

You don't need a domain admin for AAIP, you just need an account that is included in the "Administrators" group on the servers you are using AAIP on.

The only caveat is that you will want to make sure you have an open network/firewall path from your interaction proxy to your guests if UAC is turned on as UAC will not allow you to use any account other than the built-in local/domain "Administrator" SID through VIX.

[tsightler]

There are some example freeze/thaw scripts for some applications here:

https://github.com/VeeamHub/application ... 20Examples

These examples are for MySQL, Groupwise and SAP HANA, and while some of them may not be needed anymore (we have a plugin for SAP HANA now) they can still provide a good example of what is possible and how to accomplish it.

[steggles]

Thanks tsightler, those will help me out. Sorry that I hadn't replied sooner (I hadn't realised there were more replies)

I have a few questions about storage now. Currently we have two HP servers (HP ProLiant DL380 Gen9) with P411 controllers with a number of disk enclosures hanging off them acting as Veeam Backup repository servers. There are a number of RAID5 arrays setup, with NTFS formatted drives. Each disk is its own Backup Repository.

The issue is my predecessor split up a number of jobs to ensure data was written fairly evenly across all of the repositories, this has meant that he almost had a weekly job of moving VMs between jobs and Jobs between repositories to try and ensure the drives don't fill up. This is the problem I want to resolve!

I am sick of having to do this. Now, I can easily setup a scale out repository across all of these single repositories - but from what I have read I cannot move jobs from non scale out TO scale out, and is scale out the option I really want to use to solve this problem?

Would it be better to pool all of the storage across all of the drives into one Storage Pool in Windows? This would allow me to just add disks and grow the pool. But I have read that if you have a dedicated raid controller, you really should use it.

My main goal is to get Veeam to treat all of these separate arrays as one big pool (and perhaps move to ReFS to help with performance as there is a requirement to store some large jobs for 365 days)

Thanks for your help thus far!

[skrause]

The first thing I would do is run screaming from RAID-5. Especially at the disk sizes you are probably using on a backup repository.

Personally, I would go with a scale out simply because if you have issues you can contact Veeam support and get help. With a storage pool in Windows, if an edge case issue comes up who are you going to call?

[steggles]

Run screaming from RAID5 to what, RAID6? At this stage I need all the space I can get

[skrause]

For disks over 1TB, RAID5 is a recipe for disaster when a rebuild needs to happen as the rebuild times are so long and the chance for a secondary disk failure is so high. Even RAID6 can have similar issues but it at least gives you a chance to survive a second failure in the same pool during a rebuild.

A lot of enterprise storage vendors use various "extensions" to RAID to get around the rebuild time problem now as disks keep getting bigger but the read/write speeds stay the same.