Review needed on new improved setup

[tom.starren]

Hey,

Im looking at a new design for our current VEEAM environment and increase the storage as we go.

We have 2 sites with on both sites active production VMs. Each sites currently has a physical HPE server with DAS storage as proxy and a VM as management server. On each site we have a primary repository (with a retention of 7 or 14 days) and 2 copy jobs. The fist copy job creates a local archive, the second creates the same archive in the remote site (for both a retention of 4 weeks, 12 months, 1 year).

In another thread I saw that the new storage spaces aren't performing as wanted and that expanding and column sizes aren't that clear. So in the new setup I want to go with the following setup.

Site A - datacenter
HPE server with DAS storage array (60*4TB). I want to create 10 small RAID6 groups which I then add to one storage pool. In this pool I want to create several vdisk for the different types of backups. 2 NTFS volumes for the primary backups (split with a scaleout repositiory) and let Windows 2016 dedup these volumes. For the archive backups I will use ReFS volumes (for the local and remote archive).

Site B - branch office
Same setup only with 48*4TB of DAS storage and 8 small RAID6 groups.

Questions I have and hope you can help me with:
- Shall I run all the jobs from 1 management server in the datacenter and have the second management server only import the backups automatically from the repositories (through PowerShell commands) or shall I use two management servers?
- Is the current setup of 1 primary backup and 2 chained backup copy jobs preferable or should I use 2 primary jobs (1 to each site) and have a copy job chained to it for each site?
- Is the split of NTFS volumes for primary backups and ReFS for the backup copy jobs smart?

If I misted any information you would need to advice, please let me know.

Thank you.
Tom

[foggy]

- Shall I run all the jobs from 1 management server in the datacenter and have the second management server only import the backups automatically from the repositories (through PowerShell commands) or shall I use two management servers?

Single backup server is enough for this kind of setup, why do want to have two of them?

- Is the current setup of 1 primary backup and 2 chained backup copy jobs preferable or should I use 2 primary jobs (1 to each site) and have a copy job chained to it for each site?

1 primary job and two backup copy jobs for each site should be ok.

- Is the split of NTFS volumes for primary backups and ReFS for the backup copy jobs smart?

I'm not even sure you need Windows dedupe on the primary repository. I'd go with ReFS everywhere.

[tom.starren]

Single backup server is enough for this kind of setup, why do want to have two of them?

For redundancy purpose

1 primary job and two backup copy jobs for each site should be ok.

Ok, that's the way we have it now and it works ok

I'm not even sure you need Windows dedupe on the primary repository. I'd go with ReFS everywhere.

I'm afraid that the storage savings aren't as high as we currently have. I need to see how I will handle my old backup copy chains, they will expand when I move them to the ReFS volume and they now have a very high dedup ratio (110TB saved).

[foggy]

For redundancy purpose

Then you can have it standby or fire it up quickly in case of DR, import configuration and continue from there.

I'm afraid that the storage savings aren't as high as we currently have. I need to see how I will handle my old backup copy chains, they will expand when I move them to the ReFS volume and they now have a very high dedup ratio (110TB saved).

If storage savings are paramount, then dedup is justified. This is just not along the lines with our reference architecture, where it is recommended to have a faster primary storage for operational restores.

[tom.starren]

Thank you Alexander. So it would be better to give the primary storage a dedicated storage pool/RAID set?

[foggy]

This makes sense in terms of decreasing the risk of losing both local copies in case of a hardware issue.

[tom.starren]

What would you advise then?

Currently we have the following sizes in use

repository datacenter
- primary backups: 17,8TB without dedup, 4,18TB with dedup
- datacenter backup copies: 132TB without dedup, 21,9TB with dedup
- second site backup copies: 41,9TB without dedup, 3,09TB with dedup
- endpoint protection backups: 2,62TB without dedup, 705GB with dedup

repository second site
- primary backups: 9,26TB without dedup, 1,46TB with dedup
- second site backup copies: 64,4TB without dedup, 488GB with dedup
- datacenter backup copies: 126TB without dedup, 18,6TB with dedup
- endpoint protection backups: 3,92TB without dedup, 1,29TB with dedup

Both sites currently have 4 RAID5 volumes with 6x4TB SAS disk each combined via Storage Spaces. The volumes on whihc the data resides are Thin Provisioned disks on this large disk.

In the new setup I want to use Windows 2016 as said before. The datacenter would receive a one JBOD array with 60 disks (probally with the size of 4TB each). The second site would then have 48 disks of 4TB (currently 24 of them are used for the current setup).

I'm curious how you would size this, if you say the number of disks (or disk sizes) are not the correct ones, please let me know.

Thank you

[btmaus]

Hi Tom,

What setup did you go for in the end? What storage savings are you seeing?

I'm also looking to improve my storage setup for Veeam, and looking at a physical server with DAS + Server 2016 as the solution.

Keen to hear some idea's and thoughts around this.

Cheers