Snapshots are breaking bad into data protection strategies

Let’s stop writing about cloud for a moment and let’s go to the basic. As I’ve mentioned before, we can say Backup is the most challenged operation in any organization but then let me ask you: How is your current backup process? Are you happy with it? Probably you aren’t, or at least you are not full satisfied due to there is a permanent feeling that this process could be better.

Are you using tape to back-up to your most important applications?

Probably the answer is “yes”. Are you using tape as the only option to protect your data? If your answer to this last question is also positive, then probably you have the following issues:

  1. Data size matters. If you are backing-up huge amount of data, you are doing full backups weekends, and probably from Friday night to Monday morning. Daily backups are incremental (delta is not an option today) from Monday to Friday. Did you consider that the restore could last twice or more than a backup? Are you tired from waiting days to get yourdata and services online again after a crush? Does your business resist stay more than 48 hours offline?
  2. Tape usually fails. Probably you have to create more than one copy and this make worst the first issue above mentioned. The point is you never will be sure if restoring works until you got evidence through a full restore process. Sadly, you will never have enough time to try the required amount of tapes to sleep well.
  3. Data-growing is a struggle. Probably you have to add more and more and more tape drives, network devices, cards and servers every quarter or semester. Data grows twice per year, and this turn out into a nightmare when you try to keep on being using a backup performance. The bad news is, probably you can’t.
  4. Tape Costs. Despite of tape prices are cheap, the big issue is all the platform that involve to take data to tape.

Do you think to add disk to a backup process?

Well, there are reasons like speed the data transfer to back-up a high volume of files. Other reason is reliability, tape usually fails, with disk you can keep online backup copies and it’s easier to get rapid access to data and show evidence of control. Issues are not so different than tape, but reliability is:

  1. Plan Disk usage. Disk could be an expensive solution if you don’t size appropriately. Disk is expensive without the correct software to compress and de-duplicate data. Software requires memory and compute resources.  Be careful not to affect back-up timing for not sizing the correct solution. Consider data growth rates and back-up frequency to define in advance how much disk you need and how much it costs.
  2. Boost you backup on disk. Disk is faster, but you need to consider to bring enough network and compute resources to get a real advantage from this. Also, you need to think to invest in backup software,  server and agents that can manage disk natively. Take care of the whole topology from end-to-end so that you will get the strongest and the weakest point.
  3. Parallelism matters. ATA disks are good to handle sequential writing; you can get a cool performance if you send high volumes of structured data like databases directly to ATA Disks. But, usually it is good to send several streams of data from different sources simultaneously to get better performance, this helps you to get better advantage from network resources.

Snapshots are shrinking backup windows like hell

I’ve heard a lot of people stating that snapshot are not really a backup solution. And I have to say “I agree”. Snapshots are not a backup solution by itself; you need to add them to a data protection strategy to get real advantage.

I used to work at a storage company for almost seven years, I used to avoid leads based only on the raw TB cost, the best benefits of the product I was representing are pointed to the features of the software on top of the hardware and mainly how these help to theorganization to improve their RTO/RPO experiences.

Let me illustrate with this case why snaps are so important. One of the companies I had to attend, used to have a 30TB Oracle database highly transactional. They spent 4 days to send the data to tape and almost a week to fully restore it. How could you sleep at night knowing that any failure can cost almost a week to be solved and millions of dollars to the business?

The solution was that we had to use Crash Consistent Snaps. In a few words, this helps you to get a copy through a snap in no more than 10 seconds, yes anOracle 30TB ASM Disks copy in no more than 10 seconds. You don’t need agents, you don’t need to prepare database. Later, you can take the copy to other storage for better protection (now we can say it is a backup copy) and also mount it to test it –mount it and recovery can take some minutes-

Issues to Snap:

  1. Snaps reduce back-up time a lot, but not complexity. Application Administrators usually appreciate this solution more than Backup Administrators. Backup administrators usually see more complexity and they prefer to keep their tasks easier. You need to try snap copies like the ones on disk  and bring required procedures and software to warranty data consistency. You can take more frequent backup copies, but don’t abuse, because every snap need to be managed carefully as any normal backup copy.
  2. Snaps are easier to delete accidentally. Snaps are easier to take, and of course easier to delete. Again, try them like back-up copies and be careful when you clean old copies.
  3. Snaps technologies differ between vendors. There different snaps technologies, some of them requires more compute and disk capacity, others doesn’t have so much software to orchestrate and manage it, be careful, you can affect application performance and availability with a bad choice.
  4. Snaps requires additional disk capacity. Snaps disk capacity depends on technology and data change. You have to bring enough capacity to have the online copies that your business requires. Also, you can set recycle policies to save disks automatically, but again, try snaps like backup copies, because you can delete something that could avoid you a future nightmare.

Business continuity is not back-up

Don’t confuse a Data protection strategy with business continuity. Let me illustrate it with this case: you can replicate a database between two sites as part of your business continuity strategy; you replicate good data and errors. When you need to recover your data from a human mistake like an entire table deleted by accident, you need back-up.  Business continuity is required to save your neck in front of disaster rapidly, but it’s not a backup solution.

One thought on “Snapshots are breaking bad into data protection strategies

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s