back to article There's a lot more to backup than you thought

If backup is too often an afterthought, have you considered its even poorer relation, archive. You haven’t? What a surprise. The fact is, businesses and their infrastructure depend upon common services for general efficiency, management focus and cost control. All organisations do, and it would be a nightmare if they didn't. …

  1. AndrueC Silver badge
    Boffin

    It might also be worth noting that 'professional' backup software (that aimed at larger businesses) is often quite complicated.

    I'm a programmer and at my previous job we sometimes had to install and test backup software (mainly when a new version of Exchange or SharePoint was released). The infrequency meant that none of us were really familiar with the backup software packages. The result was a lot of wasted time re-learning and re-educating ourselves and all too often fighting the same battles just to get our tests performed.

    In our situation that was mostly acceptable because it was a once every year or so event. But in a proper real-world scenario you want the people responsible for backups to be familiar with the software so that they are always ready to respond in a timely fashion and always know what they are doing.

    But above all (no matter what the environment): Perform a regular test restore. Never just rely on the backup software saying the back up completed okay. Sometimes they lie. You can never really trust a backup set unless you've actually restored it to a new location and verified the data.

  2. Adam Inistrator

    nobody wants backups

    they want restores!

    1. This post has been deleted by its author

      1. Anonymous Coward
        Anonymous Coward

        Re: More on Archive

        This may sound quite daft - but two words : "Salt Mines".

        Way, way back in my personal history I worked for the owner of mega-large, mega-deep salt mines in Cheshire (three lettered UK company starting with "I" - but not either of the IT ones!). In the space where the salt used to be one found .... archived documents.

        The lesson is that the documents were sent away to a long-term storage place that was safer than houses - but that wasn't enough. You had to know what was there, how to get it back and whether you'd be able to make any sense of it.

        Apply the same lessons to no longer actively required IT data and we will get somewhere with the question of archive. Take particular notice of my statement "whether you'd be able to make any use of it". Consider whether a document created in an early word processor (what shall we say, WordPerfect, Manuscript?) being pulled from an "archived" file set. The bits in the data object might well be in perfect order - but do you have a copy of anything that can make the content readable?

        I said not to get me started ... the challenge of Archive isn't infrastructure processes (and BTW it certainly isn't storing annual full dumps on Tape - fancy reading my TOPS-20 written DUMPER format 9-track 1600bpi PE tapes from the early 1980s anyone?) - it has to be a mind set approach - how does my business preserve its business records should they be needed in the future? Having worked too in the Pharmaceutical industry I do understand the necessity of keeping near indefinite records.

        There's another approach needed where we're talking Big, BIG Data and places like CERN understand that full well.

      2. J__M__M

        Re: nobody wants backups

        That's not backup, it's redundancy.

  3. Anonymous Coward
    Anonymous Coward

    Backup ... do you mean Recovery products?

    I have spent much of the past ten years reviewing the validity of technical solutions being proposed by a major IT outsourcing services corporation.

    If I received a solution proposal describing the 'backup' solution, I immediately red-lined it and asked what their Recovery solution was. I have several classic examples of this but company confidentiality forbids me publishing them on an open forum such as this - but one of the best is proposing 'incremental for ever' (< 5% periodic churn) backup over a WAN circuit with bandwidth sized for this (typically daily) activity - and then being surprised when that they should be asked how long recovery of a total storage entity (be that file system or database) would take over the same bandwidth constrained WAN circuit. I think the lesson has now been learned ... but storage volumes continue to increase ....

    Nobody wants to perform backup ... although virtually every business wants assurance that its data is recoverable.

    (and please, don't get me started on 'Archive' ....)

    1. Julz

      Re: Backup ... do you mean Recovery products?

      Absolutely, the main component to consider is always the restore. Incremental backup's are fine and dandy, but just how long is it going to take to restore the whole caboodle? OK, so you've restored the data/system/service, what does the business need to do now? Oh, and how does this restored service play nicely and not screw up all the services that haven't just time warped into the past? These and many other potential areas for Mr Cockup all become very very important for a business that actually has to use a backup for real.

      1. John Robson Silver badge

        Re: Backup ... do you mean Recovery products?

        This is why I like the rsync and hard links option - rsync means every backup is an incremental (after the first, obviously) and the hard link architecture means that ALL of them are presented as full backups.

        http://www.mikerubel.org/computers/rsync_snapshots/

        Obviously this would need more work in a serious enterprise, but for a smaller office, or a larger home this it works very well...

        1. klaxhu

          Re: Backup ... do you mean Recovery products?

          avamar is been doing that since ..forever

    2. Anonymous Coward
      Anonymous Coward

      Re: Backup ... do you mean Recovery products?

      Just going with a project in my company that is deploying these "delta boxes", and I've felt quite lonely when I asked about the restore. Yes, the delta times are great, but... what will happen when/if you want to restore the whole lot?

      Tony, please, elaborate on "Archive"... I'm feeling a similar exercise is coming.

  4. jmaechtlen

    archives - does an archive need a backup?

    OK, maybe a dumb question - if you've archived a bunch of important stuff to tape.

    Is one set of tapes enough? What if the info is pretty important, and you need to keep it for 20 years plus?

    (or even 10 years plus?)

    1. Anonymous Coward
      Boffin

      Re: archives - does an archive need a backup?

      I think that you have answered your own question. Of course not. Where else in your IT system will you tolerate 'Single Points of Failure'? You may consider RAIT-1 or -5 (or some variant) acceptable, but in general I prefer eggs in multiple baskets.

      Also you need to consider the effective life of media, and the ability of (possibly museum quality) kit to read it in 20 years.

      In general, look back to the good practice instituted in the golden years of the mainframe and plan for (and estimate costs for) regular technology refresh and transfer of content from one generation to another of storage technology. Go read the LTO road map and remind yourself which LTO generation formats are still readable by an LTO drive you could buy today (ebay is cheating!) - and then check your media store (and salt mine)!

      (PS: where's the icon that means: "contains the flaming obvious - requires primary school level education or lower?)

  5. andysparkes

    Its good to see that restore is seen as important as backup and I really liked the comment around the restore times around incremental forever. There is only one backup appliance that I know that actually publishes predicted Restore rates as well as backup ingest rates and that is HP Storeonce. The restore capability was built into the deduping matching and store algorithms to enable that performance. As some of the previous comments suggest, its only when you go to make a claim on your insurance policy that you really know the quality of the policy you bought.

  6. Pascal Monett Silver badge

    I must say I am on the fence on this subject

    From everything I have read on backup restore solutions, nobody has ever successfully restored an entire product line.

    Oh sure, you have restores of bits and pieces. A database here, a disk directory there. Maybe even an entire server. That works, no problem.

    But then Azure goes down and we find that there is a 3-month delay to get one's web site back and no guarantee that everything will be gotten back (any word on how that finally panned out ?).

    So I have to wonder, how many companies are actually stopping their production servers to test if their restore solution is functional and reliable ? And how many have found that, yes, that is the case ?

    Just one case, please.

  7. Anonymous Coward
    Anonymous Coward

    Backup and recovery

    The thing you need to remember is that backup and recovery are not archives. Backups are to protect you from small disasters: loss of file(s), loss of a server, data corruption. It is there to get you back on your feet again. Bear in mind, I'm looking at this from a larger business scale point of view. A note on replication - if your files are corrupted or deleted, so is your replica. That doesn't help you get your needed data back...

    Archives are designed to keep data for longer periods of time. And, by their nature, the data needs to be somewhat application agnostic. If I backed up your SQL Server 2000 back in 2004 and you wanted to recover that data today, what you do? Are you going to load Win2K, apply the patches, load SQL Server 2000, apply the patches, then attempt to recover the database? What do you think the odds are of that whole process being successful?

    Cloud for backup storage? Really? In a recovery situation, can you afford the time it takes to recover your data? Can you afford the restore bandwidth? Can you afford the gigabyte charge? Cloud is fine as long as you never need to recover anything.

    The good thing about backup and recovery is that you are in control and responsible for it. The bad thing about backup and recovery is that you are in control and responsible for it.

  8. Anonymous Coward
    Anonymous Coward

    missing the point...

    I think that the IT departments needs to take a step back from storage and archive and ask some probing questions first to the rest of the business before going out to buy shiny new tin. It's estimated that 70% of data has no business value, and before you go howling that's untrue, do you think that Visio diagram or presentation you created in 2012 for a virtualisation project or monkeys nuts project is still relevant? Multiply it out to the rest of the business (who really don't understand anything about storage) and the rate of shite kept on storage is on a logarithmic scale. I knew a sales guy who backed up his photos and videos into a zip file and dumped it on a home shared drive, great idea except that he'd been doing this for 10 years and was zipping all the years of archives inside each zip file each year. A knob.

    Change tact with the approach, get some enterprise records software and tag the shite created, get the retention policies put in place so that users can't go mad and save shite forever. Pretty basic stuff... You'll find that the tin lasts longer, the backups aren't shite and that the corporate policy can be used in anger! No more PB storage of junk and having to beg/steal/borrow budgets to give away on junk storage arrays (junk cause after 3 years you get ass raped by the vendors on support and maintenance as your technology is legacy and they want you to buy the next nut sackium drive at X TB that won't work in your old array and give you £30 to trade up what you currently have that you bought 3 years ago for £3m).

  9. Nigel 9

    The below are not the same

    1. Backup

    2. Archive

    3. Archived Backups

    4. A backup of an archive

    All 4 of the above are entirely different... People need to stop mixing these up.

    Also, IMHO combining the archive and backup device for "simplicity" is nothing short of insanity....

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like