Restore Time

[steve21]

Hello Everyone Once again,

currently i am doing a test run of a restore in case of total system failure. currently i am using veeam to back up to a server and then using back-up exec to back-up veeam files to tape so i can store them off site. what i am wondering is what is a "normal" amount of time if you need to do a full system restore on environment. currently i am expecting bakup exec to take 12 hours to restore my veeam back-up files and then i will still need to restore them using veeam. is 24 hours to long of a time to do a full system restore on all my servers? all the servers combined are using about 2.5TB of raw disk space.

any feedback would be greatly appropriated.

[Vitaliy S.]

Why not to use Instant VM Recovery to bring the required VMs back to production after your restore Veeam backup files with BE? Also if you want to restore your VMs back as quickly as possible, then using HotAdd proxy servers will definitely help.

[Rumple]

is 24 hours to long of a time to do a full system restore on all my servers? - If you are talking about a complete site loss disaster then no, 24 hours is pretty damn quick for doing a restore type DR.

However if you are talking about just losing the SAN for instance...then you would not be restoring from tape and then veeam since I assume you have at least one day (hopefully 2+) stored on the Veeam backup server so you'd just start the restore right away and not have to deal with the backup exec piece at all.

[steve21]

yes i am talking about a total loss.

[tsightler]

Is 24 hours to long of a time to do a full system restore on all my servers?

I would say only you, or your really the business, can answer that question. For many companies, after a complete site loss, recovering in 24 hours is probably great, but it's really a business decision how long is acceptable.

Do you have a planned site where you will preform the recovery (another company location or co-location facility)? What I've found many times is that customers have their backups offsite, how long it will take to restore, but then make almost no plans for the steps that preceed the point of actual recovery itself, i.e. hardware, network connectivity, etc. When disaster strikes, they spend far more time getting to the point of starting the recovery than actually performing the recovery.

[nhammes]

Like Tom mentions, what are the business requirements to get back online? Do you have any SLA'a you are required to meet? Have you defined the order in which VMs need to be recovered? Do you already have Domain Controllers/Global Catalogs (if using Active Directory) running at the DR site? Can you recover only a subset of all VMs (maybe just 2 web servers instead of 4, etc) and still be able to get all services back online, therefore speeding up the recovery process... Also, again as Tom mentioned, do you actually have a DR plan/process documented for entire site failover - this requires much more than just getting VMs back online. How is DNS and routing to the new site and services going to be handled? The actual process of getting the tapes back from off-site? Personnel access for the DR site (if required)? and the list goes on... One last thought, how many VMs are you taking about having to fail over? You have to remember there are also innate limitations in the hypervisor, like the limit to the number of VMs that can be powered on concurrently, etc that will also affect failover times. In general, I think if you could recover an entire site within 24 hrs, most would consider that very good and be happy with those results, but it depends on your business. Good luck - thanks,

[homerjnick]

This question is not something Veeam can answer...

We tell our senior management team and directors that we have backups of all our data and they are happy but when we mention it would take nearly a week to restore from backup all our systems their faces drop...most non-IT staff have no concept of business continunity/disaster recovery compared to a "backup".

Therefore companies need policies to dictate how long they can go without certain systems and then IT need to plan and deploy a means to achieve those policies and targets. This is a struggle in most private based companies.

Recently we got funding to setup a DR site and using Veeam replicate all our VM's. One very important live server corrupted and sad to say the Veeam backup was also corrupted...the replica was fine and was used for over a week whilst we fixed the production server with no loss to our company and not one user noticed.

This highlighted to management that having backups (and testing them) are great but are not a replacement for any kind of BC/DR plan...needless to say more funding was granted to continue the DR plan.

So restoring from backup your entire environment should take too long...having a business continunity plan or disaster recovery plan is far more important.

[adrian.woodrup]

Lose the tapes they will cost you a lot of time and possible corruption... I would look at offsite/cloud storage. It will speed things up and can be accessed from anywhere regardless of where your business move to, you also don’t need a tape drive at the new location (if you lost the office). I would also prioritise your servers and group the ones the business deem high priority together this way you could restore them first and get the business back online with critical apps/data before waiting for the print server to restore.

[tsightler]

It's possible that offsite/cloud storage can speed things up, but if not properly planned, it's also possible that it will slow you down. When restoring lots of data, bandwidth is a major issue that has to be accounted for when using cloud/offsite storage. For example, if you have 2TB of backups to restore, and a 20Mb link, it will take ~10 days just to pull in all of the data. Not everyone has bandwidth in unlimited supply. If you store your backups to an offsite location where you plan to perform your recovery, that's the ideal solution, but if that's not possible, tapes can still be a win.

On the other hand, a truck loaded with tapes has a tremendous amount of bandwidth, it's the latency you have to plan for.

[dellock6]

You need to design a multi-tier solution: both relying on tape backups only, or offsite backup only are not good.
A multi-tier approach can help you in different scenarios, requiring each different recovery-speed/price ratio.

Luca.