Best configuration for vsphere

[willi78]

Hi all,

so as is read the forever incremental function is the best choose regarding performace and used space, correct?
for the primary job, what is the recommended number of incremental points?

for longer timespans we should use backup copy job, right?

[Regnor]

Forever incremental does not necessarily have the best performance as the merge process introduces some additional overhead. What backup repository and which Filesystem are you using? With ReFS for example it won't matter that much.

The retention is defined by your business requirements; so there's not default recommendation.
For long term retention you can use backup copy jobs with GFS to a different media. If your primary repository uses ReFS, then you can also configure GFS in your backup job; for that case you would have to go with forward incremental.

By the way; regardless of your settings, make sure that you follow the 3-2-1 rule:
https://community.veeam.com/blogs-and-p ... p-rule-569

[willi78]

HI,

we have currently about 35 VMs with a size of 3.5 TB
17 of these are critical.
2 are high critial.

Would you perform more than a daily backup for the high critical machines? (for examle 2 times a day)

And would you split the one primary job to two or three to split the machines in priority?



We have two different QNAP using ext4, in two different fire sections.

Currently on the primary job we have a daily reverse incremental with 30 points.
Then we have two Backup Copy. A longterm (12 incremental points, every 1 month) and a shortterm (copied the 30 points from the primary job)

Currently we save with a third backup copy job to an external hard disk but just every week.

So, now we think about it to change the primary job to forever incremental with 30 points.
We would not change the settings of the backup copy jobs.
and we would change the external backup to the veeam cloud connect to save it offsite in the cloud.

Is this a good plan?

[Regnor]

Would you perform more than a daily backup for the high critical machines? (for examle 2 times a day)

Again, this comes down to your requirements; and those should come from the company, departments, etc.
The keyword for this requirement is Recovery Point Objective, which means how much data loss is acceptable for a certain application.
For a static webserver this could be 24h, for a mailserver or CRM it could be as low as 1h.
So define that for your systems and then you can configure the jobs accordingly.
Here's a great post on that topic: https://community.veeam.com/blogs-and-p ... licies-699

And would you split the one primary job to two or three to split the machines in priority?

If you have 3 different RPOs for your systems, then you would also have to configure 3 different jobs.

We have two different QNAP using ext4, in two different fire sections.

Such NAS devices tend to be unreliable, especially if you use SMB. So I would suggest that you attach the QNAPs via iSCSI to your backup environment and format the repositories with ReFS. With ReFS you also get space savings via fast clone and a better synthentic performance.
If the integration is new to you, then take a look at the following link: https://www.veeam.com/blog/advanced-ref ... suite.html
If you go with SMB then at least configure health checks and/or do SureBackups.

Currently on the primary job we have a daily reverse incremental with 30 points.
Then we have two Backup Copy. A longterm (12 incremental points, every 1 month) and a shortterm (copied the 30 points from the primary job)

Currently we save with a third backup copy job to an external hard disk but just every week.

So, now we think about it to change the primary job to forever incremental with 30 points.
We would not change the settings of the backup copy jobs.
and we would change the external backup to the veeam cloud connect to save it offsite in the cloud.

I don't see much difference in doing forever incremental instead of reverse incremental. One may be a bit faster then the other, but the storage space won't change much.
But I like the idea of offloading your backups to the cloud because this will increase your data protection very much. At the moment your daily backups and your long term backups reside on the same media, which is online all the time and so an easy target for a cyberattack/trojan(just an example). So in the worst case you would have to rely on your external disk which could result in a data loss of about a week; and only if the disk and it's backups are ok. Do you have multiple disks and rotate them?
Regardless of that, yes cloud backup will be a good idea

[willi78]

Such NAS devices tend to be unreliable, especially if you use SMB. So I would suggest that you attach the QNAPs via iSCSI to your backup environment and format the repositories with ReFS. With ReFS you also get space savings via fast clone and a better synthentic performance.
If the integration is new to you, then take a look at the following link: https://www.veeam.com/blog/advanced-ref ... suite.html
If you go with SMB then at least configure health checks and/or do SureBackups.

if we connect the QNAPs via the new veeam v11 feature for a linux hardened repository then we have the same benefits and are safe regarding ransomware, right? We would use a physical linux server (Ubuntu 20.04 LTS) with XFS file system.

We would also do the offsite backup to the cloud.

[Regnor]

That's right; I forgot to mention XFS, but it does offer equal benefits to ReFS.
With the hardened Linux Repository you will be safer against ransomware. How safe depends on how you set it up; eg. single use credentials, limited permissions for the Veeam service user, disabling any remote access. Also the QNAPs need to be hardend, so that no malicious attacker can access them. If possible try to lockdown the remote access/web admin as much as possible.

But in total with the hardened repo and offsite backup (perhaps also immutable) you have a decent plan.

[willi78]

@Rognor Thank you for your feedback. Sorry I had Corona for the last couple of days.

I have more questions. We have three high critical machines. For these machines, we run incremental backups every 1 hour on weekdays from 6am to 11pm.
How often would you perform a synthetic full backup for these machines? We have a 1-5 GB change every hour.

We have all the virtual machines (about 35) divided into three RPO groups. 1 hour, 4 hours and 24 hours. Most are in the 24 hour group, only a few in the 4 hour and 1 hour group.

And one more question about deduplication. Is it more recommended to create some backup jobs to reduce the number of virtual machines in the backup job to have the ability to restore one virtual machine while other backup jobs are still running, rather than having those 3 backup jobs from the top and having to wait for them to finish? I know that the overall deduplication would not be as good in the first case.

And if I prefer the three jobs, in terms of deduplication, would it be better to split them into two more groups based on the OS to optimize deduplication, or does that have no impact?

Greets.

[Regnor]

Oh, I'm sorry to hear that; I hope you have fully recovered.

I would suggest to create a weekly synthetic full; also, if you go with immutability you need to configure those for all your jobs.

You don't need to split up the jobs. Even if the backup job is running you should be able to start a restore, as long as the job doesn't try to remove the selected restore point because if its retention. Also deduplication won't benefit from splitting up the jobs in certain groups; perhaps it would even get worse.
The only reason (in my opinion) to create separate jobs, besides the configured schedule, would be the amount of virtual machines per job or the resulting backup file size (although there's the per-vm option for that).

[willi78]

Hi @Regnor ,

yes i have fully recovered, thank you.
So we have set the retention policy to 30 days as incremental points and a weekly synthetic full.

What would you recommend for GFS? 4 daily, 4 weekly, 12 monthly?

And the next step is Veeam Cloud Connect, how many points would you recommend? also 30 daily, 4 weekly, 12 monthly synth full?

Many thanks.

[willi78]

Hi Regnor, would be great to hear from you

[Regnor]

If there are no other requirements then normally we go with the following: 14-30 daily, 4 weekly, 12 monthly and X yearly.
Cloud Connect or offsite copies would be equal to on-premises. Of you go with Cloud Connect, make sure that insider protection is enabled.