Veeam Backup schedule

[Darylrese]

Ok so I have had Veeam in place for almost a year now and have been running reverse incremental backups. I wanted as many restore points as possible on my 12TB SAN for file servers and got up to 120 points.

 

A disk failed in my SAN on Monday and now some of my backups are corrupt dispite replacing it straight away and having RAID 6 In place. Veeam are unable to recover.

 

Just about to start again but wanted to know how you run your backups?

 

If I run a reverse incremental with 28 restore points and a full active on first Saturday of each month, will it just give me a month of restore points for each copy of the full active?

 

I want a strategy that will get me several months on disk to restore files if needed.

 

I am already creating active full backup to external HDD every month for desaster recovery.

 

Many thanks

Edited July 3, 2015 by Darylrese

[Steve21]

If you're running 28 points with a monthly full, you'll only ever have those. 29 as the 28 will follow through after the full so you might have 10 full 18, or 19 full 9 etc

 

We only run 14 days incrementals (2 actives), then backupcopy those to 4 weekly, 3 monthly, 2 quarter, and 1 annual copy

 

So ends up with:

 

Last 14 days reverse incrementals

2 Active Fulls (Last 2 weeks)

Then last 4 weeks, 3 months, 2 quarters, and 1 annual fulls

 

Steve

[Darylrese]

Ok so you have a second repository?

 

If I got the NAS we spoke about, I could do the reverse incremental, 27 restore points on my NAS then backup copy it across to the NAS and start again for another month? Eventually i'd end up with months worth with every day inbetween as restore points?

 

I want a strategy that will allow me to restore on every day for several months if possible and active fulls every month for emergencies?

[Steve21]

You can do backup copies to same drive doesn't need to be different

 

Will be doing it to secondary NAS too for as you said corruption purposes, but currently doing it to same

 

I don't get what you mean by everyday for months, you'll never have daily for months unless you put your restore points up as it deletes the oldest when it hits the 28 etc

 

The backup copy does the latest version only, so the full. It won't do all 30 points etc

 

So I'll have for example (under our setup):

 

1st Jan - Annual

1st March - 1st July - Quarters

1st Aug - 1st Sep - 1st Oct - Monthly

1st 8th 15th 22nd Nov - Weekly

And then 14 days daily

 

If that makes sense?

[Darylrese]

Oh ok it only does the full .vbk?

 

Too much to get my head around on a Friday evening haha

 

Basically I just want to be able to keep as many restore points on the SAN as possible so that I can recover files from a specific date.

 

Before I had a full active then 120 reverse increments but it go corrupt so that was a bad idea as it took everything with it!

 

I just need to break it down so that if a .vbk get corrupt, I only lose a months worth for example.

 

I hope this makes sense too lol

[Steve21]

Then all you want is 120 increments, and set active fulls to month

 

So effectively you get 30reverse->Active1 30reverse->Active2 30reverse->Active3 30reserve->Active4

 

Then it'll loop, and start taking off 1, until it hits 30 and replaces Active1 again.

 

So basically 4 full backups and if one gets corrupt you still have 3 others. (+90 points)

 

Steve

[Steve21]

Active Full Backup: Veeam Backup Guide for Hyper-V

 

Pic might expain better

 

Steve

[Darylrese]

Thanks Steve.

 

That makes a lot of sense, thanks!

 

Will each full active backup take up the same amount of space? For example my file server is 1.2TB. Will each full active be 1.2TB? If so my SAN is going to run out of space before it can get to 120 restore points and 4 actives.

[Steve21]

Thanks Steve.

 

That makes a lot of sense, thanks!

 

Will each full active backup take up the same amount of space? For example my file server is 1.2TB. Will each full active be 1.2TB? If so my SAN is going to run out of space before it can get to 120 restore points and 4 actives.

 

Yep Which is why generally I was surprised you meant so many dailys. But if it's a SAN I'm assuming iscsi so do you dedup them? As that should save a lot of space as they're gonna be close files

 

I mean our fulls are 3.5TB and using 200gig~ after dedup. So my 4 weeklys and 2 monthly (21TB worth) are currently only using about 1.2TB~ (Reconfigured our backups over holidays so offloaded old ones to another drive)

 

Steve

[Darylrese]

Ok so now i've got my head around it, I think the best I could probably do is 62 restore points and full active monthly. That would give me two months on disk with two active fulls. Ill take an external active full on day 28 too so anything out of the 62 days will revert to a monthly restore and anything within the 62 days can be restored on a day to day basis.

 

The problem is because the Active Full has to be within the normal chain and these take considerably longer than the normal nightly increments, i'm going to have to try and work out a schedule that allows ample time for the active fulls to run on the last Saturday of each month but also carry out the increments overnight before school starts again the next day...that's going to be tricky.

 

Cant you select a different time for the active fulls to run on the Saturday vs what time they run for the daily backups?

 

Ahhh my head!

[howartp]

Rather than active fulls, why not use synthetic fulls?

 

Best of both worlds - a fresh start point (vbk) so your not forever relying on the initial active full, but only takes a few hours versus a few days.

[Darylrese]

Active Full only takes me a few hours per VM. Never takes days!

 

Can you separate the full actives off to separate jobs so that they can be run at different times to the incremental jobs?

[Darylrese]

How often are you guys doing a Active Full backup?

 

Two active fulls on disk of my file servers and I only have 1.6TB left out of 9TB I would need a massive SAN / NAS to get a good retention on disk.

[howartp]

My backup store is 33TB.

 

I did one active full two years ago, and synthetic fulls every Friday.

 

As long as nobody suddenly does a lot of videos one day and bumps up the backup size, I usually have 4 fulls on disk plus the intermediates between.

 

The original active full took about 4 days, but it will be nearer 5 now as we've expanded since then.

 

Peter

[Darylrese]

Hi Peter,

 

Thanks for the information. My 9TB SAN just isn't big enough to get a good schedule setup.

 

Out of curiosity, are you running a SAN or a NAS?

 

Many thanks

[howartp]

It's local storage - I bought a HP Gen8 server and specced it as high as I could get it. 14 HDDs (12 front, 2 in rear chassis). The rear 2 are raided as C: and D: (200GB and 3.5TB respectively) then the front 12 are raided as E: (33TB)

 

22GB RAM with room for more as it was planned to run the VMs in DR from it, but I'm getting a Dell VRTX this summer which will run DR instead.

 

Peter

[Steve21]

but I'm getting a Dell VRTX this summer which will run DR instead.

 

Peter

 

Would be interested to hear how that goes, as long term wise was debating getting a VRTX for our DR, as currently using Veeam server + DR host + SAN down that end. And host/SAN both need replacing soon

 

Steve

[Darylrese]

Sounds good. Unfortunately a SAN on its own (HP MSA 2040) is around £4,000 and 4TB disks several hundred each. I can't really afford that. A giant NAS is much more affordable at a couple of thousand with disks but it depends if that solution is going to be reliable and quick enough.

 

So I think I'm going to give the following a go:

 

Full Active > Reverse increments 28 days - Active Full - Reverse increments 28 days. 2nd weekend of the month full active to external media.

 

 

The problem is my schedule means I can fit in the incremental jobs during the evenings but when it comes to using the same timings for full actives, I think they are going to overlap quite a lot. I don't want to manually capture full actives every month though.

Edited July 6, 2015 by Darylrese

[Steve21]

I don't think you should worry about the timings unless you have a huge huge network or reallly slow network

 

I mean for our backups over last few weeks active full been 10-11 hours, so Sat 6pm - Sun 5am (5 TB~ moved) (Note that's 2 copies of backups running at once to 2 locations so obviously slower than normal), other backups just finish after that.

 

Steve

[Darylrese]

Thanks. A few jobs overlapping is no trouble.

 

When it takes an active full as part of the backup job, do these active fulls stay on the disk forever or are they only retained for as long as your backup chain is set to?

 

For example if I set 62 restore points and active full to once a month, will it overwrite the active fulls after 2 months?

[howartp]

If active full is part of job, it will only be retained as long as it is needed.

 

If you've set 62 restore points but never schedule another full, it will never delete the full because it's needed to feed the incrementals.

 

If you set 62 and active every month, it will delete the first active full approximately 3 months later - because ~30 of the incrementals are still based on it.

 

-----

 

NOTE: Having just written all the above, I've a feeling I'm wrong - I think it's all written based on Forward Incremental, not Reverse Incremental. I'll leave it here for anyone's future reference.

[Tall_Paul]

I used to do disk-disk-tape and followed the 'traditional' Grandfather-Father-Son backup methodology. Incremental backups Monday-Thursday with a full backup starting on Friday night.

 

It seemed to work OK. Never had a problem restoring anything (though not as easy as reverse incremental obviously).

[Darylrese]

Hmmm It seems most people are sticking with forward incremental and active or synthetic fulls.

 

I might change my backups to run in forward incremental, that way ill have a full backup each month that retains the older data that I can get it onto external HDD each month.

 

I am now trying decide between the Synology RackStation RS2414RP+ with 12 drive bays and redundant PSU's

 

 

or a 8 bay Synology Diskstation 1815+

 

 

Either solution will have WD Red 3TB or 4TB drives.

 

The NAS will take over from my very old HP MSA60 SAN.

 

NAS is considerably cheaper than SAN. I have around £2500 to spend.

Edited July 8, 2015 by Darylrese