Tape or disk backup ?

[mike.r]

I am reading the topics in this forum for a while. As far as I can see, most of the people posting here are in quite similar situations.
We all had “real” servers and did backups with Backup Exec or a similar product. We all spent plenty of money, to buy a tape or a tape library, with a range of tapes.
And over the years we where quite happy with that.

Then there came virtualization and some clever people told us, Veeam is the best product to backup virtualized Servers.
This might be right, if you backup to disks. From my point of view the Veeam disk backup strategy is smart, easy to use and quite stable.

But when it comes to tape backup we are in another world. Yes, it works…somehow…
But there are many little and bigger things, which are not so smart and bring us to many little and bigger compromises.
Most given answer in this forum “comes with next version” and sometimes with the extension “perhaps”.

This brings me to the question: Why still using tapes ?
Tapes have / had their advantages (cheap, good for long term archiving, etc.) But haven’t modern disks overrun these advantages ?
A 2TB disk is not much more expensive, than a LTO6 tape. A modern LTO6 tape changer comes more than 10.000,-- €, a modern NAS with hotswap disks 2.000,-- €.
And is it more likely that a disk don’t start spinning after 3 years in a data safe, than a tape is clotted after the same time, stored in same place ?

I think, changing from tapes to disks is just a matter of organization.
Yes, it hurts to retire a tape changer which formerly cots a lot of money.
But at latest the changer is running out of service, you have to ask the question.
And my answer then is clear.

I would really like to hear some comments to my theory in this forum.

[PTide]

Hi,

But haven’t modern disks overrun these advantages ?

Well, it depends...

cheap...

Tapes give you a price of somewhat around $0.08/GB, looks good for a massive datastore. On the other hand - prices on drives and libraries...

long term archiving

Here comes the physics - tapes are easier to transport (weight) and less mechanically sensible (Ah gravity thou art a heartless ...) + more data density per volume unit.

Also, tapes are much faster when writing/reading on/from them, however that process is linear so you'll need to wait if you want to find something.

And is it more likely that a disk don’t start spinning after 3 years in a data safe, than a tape is clotted after the same time, stored in same place ?

Can't see a reason why a tape should be clotted. The only thing I can come up with is a fire in the building so everything melts, but there can be no guaranties that the hdd will survive as well.

So, my answer is that hdds are easier to use at daily basis and they outrun tapes when it comes to random read/write operations (restoration of a latest backup, replication etc), but tapes are extremely reliable and handy when it comes to a long-term data archiving.

Thanks.

[Gostev]

Disk AND Tape backup via D2D2T is my recommendation.

The more I watch my customers and other companies handling (or failing to handle) real-world disasters, the more I like tape. It's the last resort that seems to always works - when everything else fails miserably.

In my own environment of ANY size, I would use tape and D2D2T without a shadow of the doubt.

But again, most people will probably not get it (and I was the same way too) until they spend 10 years watching tape-less companies suffer from irrecoverable data loss events, and other companies using tape to recover from absolutely horrible disasters thanks to tape alone. Yeah, of course the RTO is far from amazing - but the ability to bring the data back with close to 100% probability, no matter what happens, is what matters.

[mike.r]

Thank you very much for your answers and for showing the advantages of tape backup again.
But I am wondering a little bit, two employees of Veeam Software fighting for doing tape backup,
while their own software supports tape backup in a – how can I say it friendly – improvable way.
If you are honest, tape backup in “Veeam backup and replication” is much less usable, than disk backup.
This is one of the reasons why I wrote this article.
And, as I can see, tape backup has got no priority in Veeam software development.
The complete strategy of Veeam is made for disk backup. Tape backup is only added.
If Veeam would do tape backup better, I would love to use it again.

[vanakenarch]

I'm definitively with Gostev for D2D2T, but Veeam has to make the backup to tape more stable and reliable then it is now.
I still use tape as an easy way to bring your backups off premises (yes I know you can use Cloud services, but I don't want to).

[martynuk]

Veeam already have some tape improvements in the pipeline for v9:

http://www.veeam.com/blog/v9-robo-and-t ... ments.html

There is still room for further improvement, but I'm sure they realise that. They will also take on board any suggestions made through this forum, and if enough people ask for the same thing it could happen.

[Gostev]

And, as I can see, tape backup has got no priority in Veeam software development.

And yet somehow tape is by far the biggest section of the What's New in v8 document

[mike.r]

And yet somehow tape still does not work stable in a business critical environment

[dellock6]

With all the latest news about cryptolocker and other malware spreading around, tapes have another advantage that I would not underestimate: they are the only really viable way to have a proper offline copy of your data that cannot be touched once stored. Anything that is connected by network to a system can be reached, but only physical intrusion can access a tape.
And tapes still have a physical switch to put them in read only mode...

[Dima P.]

Hello mike.r,

Can you please refer to some particular functionally you are missing, issue you are facing or problem you’ve reported to support team? Don’t get me wrong: you keep saying our tape support does not suite your needs but does not refer to any specific example of your personal use case or critical functionality we should implement.

I’ll be glad to hear and discuss any feature request with you and we come to the conclusion its indeed valid – it will be implemented as soon as we get available resources from R&D. By the way that does not refer somehow to low tape feature priorities: we have a dedicated tape support team

[mike.r]

Hello Dima P,
please don’t get me wrong. I am not here just to crap. I use Veeam every day in a production environment and I love the features Veeam got when I backup and restore to / from disk.
But read the topics in this forum by yourself. There are some major topics, which make your product hard to use.

Example 1 (Drive cleaning)
A topic, open for two years now, 3384 views. You think this has a “low tape feature prioritie(s)” ?
Have you ever made a backup on 7 tapes and after 30 hours of work, when changing from tape 6 to 7, your complete work is destroyed by stopping backup for a simple drive cleaning ?

Example 2 (Backup to Tape only last backup with reversed incremental)
A topic, open for two years now, 16701 views. As long, as backup to tape is so hard to use as it is, it would be very helpful, if you just backup the files you really want to / have to backup. (s.a.)

Example 3 (Recovery from Tape - Needs major work)
A topic, open for one year, 3212 views. And it does’t need only major work, it does need major time too. You think it is not so important to recover in a reasonable time from a tape ?
Just read what your customers think.

These are just three examples of major problems and none of them is marked as “solved with V9”.
By the way: The sometimes mentioned article “v9-robo-and-tape-backup-enterprise-enhancements” does contain only one new tape feature.
This is really everything you have done for tape users in V9 ?

[Dima P.]

please don’t get me wrong. I am not here just to crap. I use Veeam every day in a production environment and I love the features Veeam got when I backup and restore to / from disk.
But read the topics in this forum by yourself. There are some major topics, which make your product hard to use.

I respect and highly appreciate any feedback, though, detailed feature request supported by a valid use case makes more sense and, eventually, helps a lot to set the right priority and push this feature into development.

Example 1 (Drive cleaning)

While I agree this feature is important the only reason it’s postponed because of the available workaround to use libraries built in cleaning logic.

Example 2 (Backup to Tape only last backup with reversed incremental)

As stated in the thread there is (and was) an existing workaround to use a reversed incremental backup mode, manly, this topic is popular because of the various proposed methods of achieving this goal. solutions.
We recognized that the proposed workarounds are complex for all new comers, so in the latest update we added the ability to chunk the backup chain and start the backup from the latest full. “Additional care” for the rest of the cases will be provided in the upcoming major release as stated in the thread.

Example 3 (Recovery from Tape - Needs major work)…. You think it is not so important to recover in a reasonable time from a tape ?

I’ve actually asked users explain which type of restore should be provided and why in order to so set the right priority for the next major version. So consider this feature request as pending because of the open discussions

These are just three examples of major problems and none of them is marked as “solved with V9”.

Two.

By the way: The sometimes mentioned article “v9-robo-and-tape-backup-enterprise-enhancements” does contain only one new tape feature.
This is really everything you have done for tape users in V9 ?

Four major features were discussed in the announcement: Global Media Pools and library failover (i.e. unlinking the media pool from the library), Parallel Processing between drives, GFS Media pools and Unknown Medium Changer support, however “small” features are still hidden (possibly, till the public beta or even release)