IT Employment

Ask something
Discussion Locked

Backup Criteria

0 helpful
12 replies

by kylegeldmaker 2 views

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Hello,
I was wondering what the common, and not so common IT Professional demands of his/her backups. I have been given the task of finding the best way to backup our data. I already do tape backups, I am looking for ideas on geophysical redundancy, and offsite backup storage. What features are there? What features should there be? What would be the most Ideal backup situation? Be creative, think outside the box on this one. Thanks a million.

12 replies

by faradhi

Backups are fitted to match the Disaster Recovery plan. So the best solution for backups will vary depending on the business.

A high volume business such as banks and financial instutions will want offsite replication in a diffrent geographic area. So will businesses that work over a large geographic area.

Low volume businesses in one city will be fine with tape.

In summary, The IDEAL backup solution will depend on the disaster recovery plan. This will differ from business to business.

by kylegeldmaker

Ok, my company is of decent size, the larger side of decent actually. What I want to know now is what features should an offisite location have? What are the most common features? Which should I look for? Which should I stay away from? I have been doing my own research on this and I have found that most of the places that offer services close to what I am looking for are mom and pop type places. Each offer something different.

by hal 9000

With an impeccable track record for any form of storage so that immediately rules out the small operators.

Again it all depends on the company and the data that you need to backup generally speaking the large Multi nationals don't outsource their data they just replicate it in different parts of the world that are considered as Friendly and Safe. Like for argument sake say you are a company like Intel you wouldn't consider using a storage facility in Afghanistan to securely store your data would you?

You would replicate this data to different parts of the country depending on where your head office is and just how potentially dangerous it is. Like for argument sake if you where on the wrong side of the fault line in CA you would move all your critical data to the East Coast and even then you would also replicate it in other areas where you had major infrastructure so as to reduce the possibility of total loss in the event of some form of Cataclysmic event happening. It could be something as simple as something like [b]3 Mile Island[/b] going the wrong way that would make everything within several thousand miles uninhabitable so everything within that area would be disposable to the company.

Or if there was a concerted Terrorist Attack against the Government the easiest way for domestic people to take down the Government would be a wide spread and protracted attack on the Power Distribution Grid even a Natural Disaster could manage this if it was widespread enough.

You could most likely survive a week without electricity but any longer and you are going to see a complete collapse of the society no electricity means no water and no way to pump petrol or for that matter crack oil into Fuel so the longer the event was to continue the worse the position would be. Of course without electricity long term you would not have the ability to use banks or refill shops so you would have major food shortages and the inability of people to move to a different location where things where safer.

I've yet to see an DR Plan outside of Government address issues like this and even the Government DR Plans are very limited in their scope not to mention horrendously costly to implement and keep active till required.

So just how severe are you planning for and how solid does your DR Plan need to be?

Col

by tonythetiger

We have set up an identical server (Raid 1 OS, Raid 5 Data) to our main file server in another building in our district complex(connected via fiber). Nightly, we "robocopy" the data drive on our main file server to the other one. (in case of main server crash, we can be up in minutes with at worst day-old data. Also, since most of our restore requests are for files from the previous day, dragging and dropping from the backup server is faster than restoring from backup). Just prior to this robocopy, we backup the 'backup server' to a USB hard drive, which we then send alternating disks to one of two offsite facilities (connected via T1). We used USB because every workstation in the offsite facilities has USB and it's a simple matter to call them and have them hook up if we need to restore from one of them (formerly we had to drive to get the tapes).

We back up incrementally to three alternating disks nightly average about 4 gig) and do fulls once a month to 6 disks (average 350 gig) (each disk will hold two full backups). We can thus go back a full year. Despite best efforts the data continues to increase though, so we'll likely soon be going to one disk per full backup.

We have another file server but it is reserved for static files (files needed online for reference but seldom if ever changed). These files exist on other media (CD, DVD) that are stored in a fireproof safe, so backing up is not really necessary. However we do back it up to tape occasionally (a couple times a year) and send the tapes to the same offsite locations.

We've been doing this since last October and it's worked well for us. We've had one drive go flaky (I was able to copy the data to another drive, so no data loss) and it was replaced under warranty. Restores, even across the T1, are far faster than tape for one to several hundred files because of the random access nature of disk.

by tonythetiger

We have set up an identical server (Raid 1 OS, Raid 5 Data) to our main file server in another building in our district complex(connected via fiber). Nightly, we "robocopy" the data drive on our main file server to the other one. (in case of main server crash, we can be up in minutes with at worst day-old data. Also, since most of our restore requests are for files from the previous day, dragging and dropping from the backup server is faster than restoring from backup). Just prior to this robocopy, we backup the 'backup server' to a USB hard drive, which we then send alternating disks to one of two offsite facilities (connected via T1). We used USB because every workstation in the offsite facilities has USB and it's a simple matter to call them and have them hook up if we need to restore from one of them (formerly we had to drive to get the tapes).

We back up incrementally to three alternating disks nightly average about 4 gig) and do fulls once a month to 6 disks (average 350 gig) (each disk will hold two full backups). We can thus go back a full year. Despite best efforts the data continues to increase though, so we'll likely soon be going to one disk per full backup. It is faster than tape to backup. We were taking about 56 hours to do a full (Data drive plus C: drives of other servers). Now it takes 21 hours.

We have another file server but it is reserved for static files (files needed online for reference but seldom if ever changed). These files exist on other media (CD, DVD) that are stored in a fireproof safe, so backing up is not really necessary. However we do back it up to tape occasionally (a couple times a year) and send the tapes to the same offsite locations.

We've been doing this since last October and it's worked well for us. We've had one drive go flaky (I was able to copy the data to another drive, so no data loss) and it was replaced under warranty. Restores, even across the T1, are far faster than tape for one to several hundred files because of the random access nature of disk (plus we save the time it takes to drive to get the tapes).

When including the cost of the tape drive, it was also less expensive.

by kylegeldmaker

Is is possible to set something like this up without having RAID. I was thinking like having a T1 connect to a remote location and dump the server data on the computer without the computer having to be a server or something like that. Maybe instead of a computer just having a network drive, would that work?
Also is it possible to back up just the deltas?
What is the typical size of a backup? I haven't really looked. Could a backup be written to a DVD?

by tonythetiger

The backup disks aren't RAID, they're your garden-variety USB external hard drives.

A couple gig would probably be OK across a t-1 (I have restored that much) but several dozens or hundreds of gig would be out of the question.

by kylegeldmaker

Ahh, Ok I see.
I am not sure how large our files would be. we have 2 file and print servers at this location. both have 4 74 gig drives that are mirrored. I would assume that the backups would be >10 gigs. That should be fine, right?

Also, are you encrypting the transfer? If so, what would be a good encryption method?

by hal 9000

the encryption method is up to you but you would need to store the encryption Key somewhere else or even with the data so you could rebuild if there was an event of sufficient scale to destroy your current infrastructure. Floods are really good at doing things like this as it can take weeks before you can gain access to your hardware again and even then it's generally garbage as well so you're carefully encrypted data would be unreadable once you got it back as the Encryption Keys no longer exist.

Honestly there are so many possibilities with [b]Natural Disasters or Acts of God[/b] that can destroy an entire City or even all of the coast of a country. A Tsunami could be generated and wipe out the entire east or west coast of a country so everything affected would be gone or at the very least destroyed. If something like that was to happen then it all depends where you are as to how you'll be affected. If you happen to be on one of the islands in the middle of the ocean nothing is considered as safe on that piece of dirt. If you happen to be in a major country then your remote storage would need to be as far inland as possible so as to avoid any damage though the power supply may be interrupted but at least your data would remain intact.

Having redundancy built into a single building will save you from normal everyday events but will do nothing for the out of the ordinary events.

At one place recently during heavy rain for a very short period of time [i]30 minutes[/i] there was localised flooding which flooded the server rooms of several major buildings destroying all the hardware contained in them.

Then you have the [b]Once in a 100 Year Events[/b] to look at as well as these do happen and it doesn't matter if they haven't occurred for 300 years that just means that it's more likely to happen sooner rather than latter. If you are really serious in implementing a really decent DR Plan you never have your server rooms below ground level you actually need them a few floors up to prevent flooding or water penetration but this isn't normally acceptable to business so they bung their server rooms below ground and suffer the consequences on an irregular basis.

Do you live in an are that is Earthquake Prone? If so you need to have all your current data replicated a long distance away so as to be unaffected by any earthquake. It's possible for a severe Earthquake to affect the water storage facilities for major cities so you could get a double whammy effect first the earthquake partially destroying buildings and then having to evacuate the entire place just in case the dam collapses and if it does then everything will be washed away by the [b]Wall of Water[/b] that is the result of the Dam collapsing.

Now are you sorry that you said to think outside the Square? :D

Col ]:)

by tonythetiger

Not sure what kind of encription they use, but it is encrypted. Additionally, the backups are done on an isolated infrastructure we own and have exclusive access to. The restores done from the remote sites are across commercial T-1 (actually part of one. Two sites per T-1 voice and data), but since the data is already encrypted, and doesn't get unencrypted until it reaches the server, there is still some safety. Plus, I've only done about 4 restores in the last 6 months from the remote site. Most of them are the previous day's which are on the other server.

My rough guess is that 10 gig would take about 14 hours to back up across a T-1.

by hal 9000

Depends on what it is you are backing up. If they are company records they do not need to be treated as safely as Medical Records but that doesn't mean to imply that you can be sloppy either.

I personally prefer tape but that's me and mainly because of the people that I work for it's necessary. But there are numerous other Backup solutions available depending on your individual needs ranging from backing up across a WAN to several different locations in one country to world wide.

If you don't have this option you need some safe Off Site Storage facility that is both Fire and water proof in the case of a major catastrophic event occurring.

But again it all depends on the business as if this is a small business a major catastrophe may well render it dead if all the people working there are killed or severely injured at the same time or by the same event. After that it's only the Tax People who would be interested and really they wouldn't be considered as overly important by those left behind if there where actually any survivors. So having an off Site Secure Backup facility in a situation like that is needless.

If you have a medium to large business but still basically a local enterprise then you really need to think about some form of Secure Off Site Storage that is both accessible and secure preferably a long way away but close enough to easily get your data back when required but far enough away not to be affected by some local Cataclysmic event like an Earthquake or something similar.

Really as far as I'm concerned this is all Academic as unless you know what it is you are supposed to be protecting and who it is being protected for everything is just best guess and that's all that can be expected.

Where do you draw the line here for securing your data and being able to rebuild the business?

After a Nuclear attack or a Dam failure that washes away everything in it's path or what?

Do you want to protect this data from a Accident at a facility the deals in fissionable materials that will contaminate a wide ares for hundreds of thousands of years?

While it can be done is it worth the effort and cost?

Col

This discussion is closed to new replies.