What next after LTO ?

Caporegime
Joined
18 Oct 2002
Posts
33,438
Location
West Yorks
I have a friend who runs a small business with his family. Their backups are now becoming so large that an LTO drive is not enough to backup their 3 servers.

What's the next step up. I work for an IT service provider that hosts over half the UK's medical records so our backup solution is a massive tape auto loader that takes up 5 racks !

I've spent the night searching and most Backup to disk devices or tape auto loaders are just massively expensive. When you need to backup more than 2.5 bt a night where do you go next ?
 
First question is what on earth are they doing to be generating 2.5TB of changes per day. Secondly Tandberg make autoloaders that are a bargain.

If you want to move away from tape then look at DataDomain or one of the Dell AppAssure appliances.
 
First question is what on earth are they doing to be generating 2.5TB of changes per day. Secondly Tandberg make autoloaders that are a bargain.

If you want to move away from tape then look at DataDomain or one of the Dell AppAssure appliances.

You need more than just the changes on tape though. The whole data set gets backed up then you do an incremental backup to keep it up to date.

How would you restore your servers from tape if the tape only contained the incremental updates ? You'd need another tape with the first initial data set on and that's where we are falling over. The point here is we're backing up the entire server so it can easily be restored as a VM so just backing up exchange data is not what were after.

What we'd like is something that would enable us to restore any of the three servers quickly and easily and preferably something that would enable offsite replication for DR recovery.

The Dell app assure stuff loks like what we need, but I read one of the devices had a list price of £15,000 which is too much for our budget
 
How are you backing up the VM's at the moment? People can advise better if they know the software you are currently using. You also haven't stated what virtualisation technology you are using.

Just putting in a second tape when the first is full is the simplest way for backing up your fulls, then have an incremental job run for the remaining days of the week.

If it's a small business without a huge chunk of money and it's all VM's running on VMWare, then I would look at either Veeam or vRanger. Both can do backups and incrementals of VM's (to disk), to be backed up to tape by whatever your solution is.
They also do timed replication, which would potentially satisfy your DR requirement too. Just don't ever believe that replication is a backup and you'll be fine.
 
How would you restore your servers from tape if the tape only contained the incremental updates ?

My mistake, I read it as you were struggling to backup 2.5TB every night, not that you'd hit that limit on some of your jobs.

Do you have a budget in mind at all? A Tandberg LTO6 autoloader is £2700 plus VAT at retail, I'd expect to see a fair chunk off that just for asking.
 
Without knowing too much about why you have so much data to backup, if I understood correctly, I would split the backups:

- Backup to disk using something like Veeam or vRanger (I have no experience with any of these, but some of them are highly rated). Because they do compression and de-duplication, you wouldn't need a huge array, so maybe something with 3 x 3TB SATA disks in RAID 5 for a total of 6 TB storage?

- Backup critical data to tape (i.e. specific directories on the VMs)

The backup to disk will give you quick recovery in case of some sort of disaster (you really have to define what these disasters are, as you may find that just backing up doesn't cover a lot of scenarios), and the data backup to tape will give you archival long-term storage and recovery of what's important; you can always rebuild a VM in an emergency -- it's the data that matters.

If you have hosted Exchange, consider going Office365 -- this will significantly cut down management and backup requirements.
 
Without knowing too much about why you have so much data to backup, if I understood correctly, I would split the backups:

- Backup to disk using something like Veeam or vRanger (I have no experience with any of these, but some of them are highly rated). Because they do compression and de-duplication, you wouldn't need a huge array, so maybe something with 3 x 3TB SATA disks in RAID 5 for a total of 6 TB storage?

- Backup critical data to tape (i.e. specific directories on the VMs)

The backup to disk will give you quick recovery in case of some sort of disaster (you really have to define what these disasters are, as you may find that just backing up doesn't cover a lot of scenarios), and the data backup to tape will give you archival long-term storage and recovery of what's important; you can always rebuild a VM in an emergency -- it's the data that matters.

If you have hosted Exchange, consider going Office365 -- this will significantly cut down management and backup requirements.

This is kind of what I was thinking

Backup to disk somewhere for quick retrieval in the event of a data corruption / faults on one of your servers.

Then backup that disk backup to tape and put the tapes in a rotation meaning you can take a tape offsite everyday to provide you with a basic disaster recovery in the event of a fire.

My mistake, I read it as you were struggling to backup 2.5TB every night, not that you'd hit that limit on some of your jobs.

Do you have a budget in mind at all? A Tandberg LTO6 autoloader is £2700 plus VAT at retail, I'd expect to see a fair chunk off that just for asking.

Thats another option that I was looking at, but the Dell ones were silly money - starting from 5k. The tandberg sounds much more affordable.
 
just so you know the

HP 2024/4048
IBM TS3100/3200
Dell TL2000/4000
Overland Neo200/400S

are all the same library, except one holds 24 tapes and the other 48 so for future repair these will be cheaper, tandberg on the other hand are a similar design but not the same therefore less common. just saying. ide go for one of the others.
 
Last edited:
I wouldn't spend a penny more on tape drives or libraries. Backing up entire VMs is not the same thing as backing up data.

Those Dell AppSure look really interesting. Let us know if you get prices...
 
Tape doesn't care what you back up onto it. I agree that if you are running VMs that you should be backing up at the VM level and not the guest level, but Veeam will happily talk to tape libraries.

No matter how you look at it, it's still the cheapest way of getting large amounts of offline backup, at the expense of recovery times.
 
It'd also be handy to know what you're backing up: DB's, File shares, Mail?

For our fileshares we use vRanger to backup the OS, and commvault (Though any agent based backup software would do) to backup the data itself.
Full image backups are great for DR, not so granular when it comes to when the director has deleted a single file that they absolutely must have back right now.

There are lots of ways to do backups, and having everything as a VM gives you a lot more versatility in how you achieve it.
 
It'd also be handy to know what you're backing up: DB's, File shares, Mail?

For our fileshares we use vRanger to backup the OS, and commvault (Though any agent based backup software would do) to backup the data itself.
Full image backups are great for DR, not so granular when it comes to when the director has deleted a single file that they absolutely must have back right now.

There are lots of ways to do backups, and having everything as a VM gives you a lot more versatility in how you achieve it.

All of the above. so interesting question for you then.

If you were to start creating a backup solution from scratch, that enabled backup of both the data and the VMs for disaster recovery, what would you do ? What hardware are you backing up the vRanger / Commmvault backups on ?
 
You'll need to look further into licensing. There was a minimum requirement of vSphere essentials for backups, and there were some references in release notes to this no longer being a constraint. So you may still need it, you may not it'll depend on your vSphere version.

Software
In your situation, I'd use vRanger or Veeam for DR, timed replication can be setup for your VM's depending on rate of change and available bandwidth to a similarly spec'ed ESX host.

I'd also use it for image based backups of your VM's, with some additional software to backup to tape (Arcserve, Acronis, whatever's your poison). Veeam can write to tape too, but I don't think it'll help if you want to do agent level backups - someone else may be able to comment there, as I've not used it in years.

You know your customer/friend better, but whether you use agent level backups for exchange/SQL/File is over to you. Image backup is great for DR or where you'll only ever have to restore an entire disk or the entire server, but it's unwieldy for granular restores.
Cataloging in vRanger (I presume Veeam has something similar), can speed up file level restores but I've not tested how good it is, so you'll need to get a trial and test for yourself if it's restore times are acceptable to the customer.

Aaaaand finally, assuming you have AD, don't take an image backup of a DC - use the built-in tools MS recommend. If you have just 1, you might get away with it, more than that and you are heading for a world of pain.

Hardware
Rotor is spot on with his recommendation, something with a nice chunk of storage and the tape drive attached. You'll have vRanger/Veeam and your tape software installed on it - not a great deal more to say about it.

With the compression that vRanger/Veeam provides you probably won't need the full 2.5TB LTO6 can provide. But even if you do, as long as the job timeouts are set sufficiently someone just needs to change the tape when they get in and let the job complete.
This will only be once a week for your full, every other day should happily fit on a single tape.


We backup to LTO3/4 tapes in an HP MSL8096 tape library. We have seperate servers running vRanger and the tape library, but it's otherwise not dissimilar to the setup outlined above.
 
Modern backup software that is designed to work with VMs is more than capable of granular restores. Veeam supports individual Exchange objects, SharePoint and SQL as well as individual files.

I'd highly recommend a chunk of RAID6 storage to store a week of backups on before shipping them off to tape though, just to make restores less likely to induce suicide.
 
We use something similar to Little Crow.
vRanger to backup the VMs to MSA2000 array.
Backup Exec 2012 to backup Exchange 2010 & other servers.
Disk to Disk/Disk to Tape from the backup server.
Enough Disk space on the SAN for a Full backup & a weeks worth of differentials.
Makes doing restores so much easier.
vRanger is well worth the money, you can do selective file restore from backup images.
 
Caged is right, as long as the user has told you while the backup data you need is still on disk. We keep a week, and this is generally enough, but YMMV.

In a worst case (The repository has already cleared for the next cycle, then they tell you about the file), then you have to restore from tape, and if it's an image that means the whole disk/server.
So in addition to the hardware for backup, you'll need to make sure you have capacity for restores when needed.
 
Back
Top Bottom