Advise: Hourly Forward Incrementals to CC

[CloudMSP]

This is what we are currently trying to implement at every one of our MSP clients.

1. Dell T20 (E3 proc) with 2 or 4 WD RED NAS Drives in MIRROR mode with Windows Storage Spaces as the BR server at each client location.
2. Hourly forward forever incremental backups from 6 AM to 7 PM M-F. Retain 120 restore points locally (about two weeks).
3. Seeded to our Cloud Connect Server NIGHTLY keeping the last 7 restore points (7 days).

It seems to work so far but at some clients I'm seeing merging take too long sometimes, sometimes the merge takes more than an hour so it misses its next hourly incremental.

Another issue I'm seeing right now is that a merge is now going on 30+ hours at 63% (it was previously interrupted if that matters).

I just realized that in the first issue the backup is actually going to a Synology doing CIFS and the second issue its to an attached USB 2 device (G Data). Two separate customer and issues.

Ultimately I'm asking is if I stick to the 1,2,3 I've outlined above, is this the right way to go about it? Do I need to still do active fulls or other backups to prevent corruption. Is my setup good enough for synthetic fulls, or is that even a good idea? I was reading in other threats that you might want to run an active full periodically because the chain can get corrupt , but how does that help the cloudconnect chain, it would still be going forever inc based on the seed right?

Please help. Thanks.

[foggy]

I just realized that in the first issue the backup is actually going to a Synology doing CIFS and the second issue its to an attached USB 2 device (G Data). Two separate customer and issues.

Both are not quite fast repositories in terms of I/O, have you tried to back up to some ordinary Windows box for performance test?

Ultimately I'm asking is if I stick to the 1,2,3 I've outlined above, is this the right way to go about it? Do I need to still do active fulls or other backups to prevent corruption. Is my setup good enough for synthetic fulls, or is that even a good idea? I was reading in other threats that you might want to run an active full periodically because the chain can get corrupt , but how does that help the cloudconnect chain, it would still be going forever inc based on the seed right?

Here's a good thread on this matter, please review.

[CloudMSP]

Ultimately I'm asking is if I stick to the 1,2,3 I've outlined above, is this the right way to go about it? Do I need to still do active fulls or other backups to prevent corruption. Is my setup good enough for synthetic fulls, or is that even a good idea? I was reading in other threats that you might want to run an active full periodically because the chain can get corrupt , but how does that help the cloudconnect chain, it would still be going forever inc based on the seed right?
Here's a good thread on this matter, please review.

I just read the threads thanks, but still I need guidance. My chains are 120+ long because I want two weeks of hourly backups on my local BDR forward forever. Nightly they are sent to cloudconnect to create a chain of 7 dailies.

If this 120 chain gets corrupted, my offsite will also be corrupt?
Does the cloudconnect ever need re-seeding?
Should surebackup be ran on the cloudconnect side as well ? (doesn't seem feasible for me)
With surebackup a 120+ chain should be no worry at all?

It's a big pain to seed the backups to CC if that's ever required, that is a problem for me.

[foggy]

If you use SureBackup to make sure your local backups are recoverable and health checks on the backup copy jobs to the cloud repository, you can be sure they are also fine.

[CloudMSP]

How does SureBackup really ensure the back is good? Especially when talking about a large multi terabyte file server? Just because it boots doesn't mean its good. From other posts on this forum it seems like the answer is, it can't really.

Is the "Validate entire virtual disk contents" option something that should always be used?

My backups run hourly until 7 PM nightly, I should then run a SureBackup w/ Validation before starting the offsite BCJ? Is there a way to link this so the SB runs the BCJ directly after it's completed?

SB all seems very complex just to ensure you have a valid backup - and I've got to do this across 50+ customers. Is Veeam really the right product for a MSP who needs to scale ?

[foggy]

How does SureBackup really ensure the back is good? Especially when talking about a large multi terabyte file server? Just because it boots doesn't mean its good.

Is the "Validate entire virtual disk contents" option something that should always be used?

Backup file validation is an additional check that allows to make sure the file itself was not altered after being stored (due to storage corruption). Along with the boot check and application tests it ensures the VM will recover and act normally.

My backups run hourly until 7 PM nightly, I should then run a SureBackup w/ Validation before starting the offsite BCJ? Is there a way to link this so the SB runs the BCJ directly after it's completed?

SureBackup should be run on the primary backups, backup copy has built-in mechanisms to ensure the target backup is bit identical to the source.

SB all seems very complex just to ensure you have a valid backup

It's not "just", it is all about that. You do not want to find your (or your customers') backups not recoverable in case of DR, right? While it might look complex, it is set once and then run automatically.

[CloudMSP]

Ok thanks, can you confirm that a 124 hourly restore point chain is a bad idea and prone to corruption?

How can this be mitigated ? Would it be a GOOD or PIRATICAL idea to have two chains?

1st Chain: 124 Business Hourly Forever Incremental Restore Points (About 2 Weeks)
2nd Chain: 30 Daily Restore Points Forever Inc (About 1 month) -> Backup Copy Job to Offsite Nightly (7 Restore Point Retention).

Both chains would be on the same storage (T20 Server, WD NAS Drives Storage Spaces Mirror Array).

Is two chains a good solution to my problem, even if it's on the same storage? Please let me know the truth because this would be a lot of work and disk cost to do this everywhere. Is a 30 long forever inc reliable and recommended?

[CloudMSP]

Please, any one? Should I be asking these questions directly to support? I thought this forum would be a better place to really get some questions answered but I wonder if I'm in the wrong place.

[foggy]

The chain consisting of 124 restore points is ok provided you do at least periodic recovery checks. Even if you go with the two jobs scenario, recoverability checks are still required.