Off Site Backups - tape or HDD?

Haggisman

Haggisman

Haggisman

Soldato
Joined
6 Oct 2004
Posts
19,329
Location
Birmingham
So having recently taken over the position of IT everything in a medium sized business, I've been tasked with implementing an offsite backup solution, since my predecessor didn't have one in place.

Our current backups take ~4TB, however chances are this is going to increase, also, our internet upload speed is ~1.3Mb, so those factors pretty much rule out online backups.

I've been looking at a couple of solutions involving both removable HDDs (USB3) or LTO4 tapes (SAS), but I will confess I'm struggling to decide which is best.

So far the systems I've been looking at are:

Buffalo Drivestation Quad USB3,
8x 2TB HDD (2 sets of 4 to be rotated weekly)
- Approx total cost £850-900

or

Quantum SuperLoader 3 LTO-4HH tape drive
10x LTO4 Ultrium 800GB/1.6TB tape cartridge
Cleaning cartridge
PCIe SAS card
SAS cable
- Approx total cost £2700

I'm leaning towards HDD myself, purely based on ease of use and cost, but suppliers who have been giving me quotes so far have been insisting tape is far better - however I'm somewhat suspicious that they are slightly biased based on cost ;)

Just wanted some independent recommendations or ideas really - or if anyone has been in a similar position and what system they put in place?

Many thanks :)
 
We use a Disk to Disk to Tape solution, which envolves an array of disks attached to our backup server, backup exec runs every night, pulls all the data to the disk array, as soon as the job finishes a duplicate job runs which then backs up from the array to an LTO4 tape library for offsite storage.

Works well.
 
My concern with trying something like the Buffalo would be how do you transport the HDs around to ensure they are protected? An LTO tape is far more robust for taking off-site on a regular basis - HDs are going to need careful packaging and transportation.

How does the DriveStation handle the HDs being changed weekly? It won't be much fun if each HD swap means you have to re-initialise or integrity check the array...
 
We use tapes at work, I see a stupid amount of them every day, they work and are reliable...but the drives seem to be older than me! Dont know if this is just MOD type standard as they are always behind.
 
We use tapes, I think all I know is the IT guy drops off a big red case, then some company picks them up and leaves an empty case. Surely everything is backed up on the servers anyway? how likely is it that all 4 servers fail at once?
 
Jay794 said:
We use tapes, I think all I know is the IT guy drops off a big red case, then some company picks them up and leaves an empty case. Surely everything is backed up on the servers anyway? how likely is it that all 4 servers fail at once?
Click to expand...

Depends how much of a pyromaniac you're feeling today? :D
 
ruffneck said:
We use a Disk to Disk to Tape solution, which envolves an array of disks attached to our backup server, backup exec runs every night, pulls all the data to the disk array, as soon as the job finishes a duplicate job runs which then backs up from the array to an LTO4 tape library for offsite storage.

Works well.
Click to expand...

This is similar to what I'm looking at - currently backup exec pulls everything to a NAS. My concern is the offsite part :)

#Chri5#, that's a good point about transportation and a major point I'll have to consider, I know that tapes are a lot more robust - I guess my main concern at this point is the cost!

Jay794 - it's very likely if the building catches fire :p

Thanks for the replies guys, I'll do a bit more digging around the tape side. :)
 
i used lto 4 tapes and wont change.

I use 15 tapes nightly over 3 servers (2 windows 1 oracle) and all tapes are stored in a fire proof safe, rated to over 1600 degrees!!

So no need to take tapes home.

I also use the 2008 r2 feature 'windows server backup' to create nightly images of servers in case the OS fails. very handy. (saved my ass only last week!)
 
ruffneck said:
We use a Disk to Disk to Tape solution, which envolves an array of disks attached to our backup server, backup exec runs every night, pulls all the data to the disk array, as soon as the job finishes a duplicate job runs which then backs up from the array to an LTO4 tape library for offsite storage.

Works well.
Click to expand...

Replace backupexec with netbackup and this is what we do :)
 
#Chri5# said:
Fire, flood, theft...
Click to expand...

Theft? hardly, the servers are the size of fridges and the facilty has security on site 24/7. Fire? possibly except the server room has one of those vacuum systems to remove all the oxygen from the room, thus stopping the fire. Flood? possibly, I suppose if it rained hard enough the upstairs could flood but the roof would have to cave in first
 
we use netbackup, daily backups run mon - thurs, weekly backups running friday night for 13 weeks, monthly backups which run monthly and 7 years with of yearly tapes.
 
I like the Disk to Disk to Tape solutions myself and use to support one running Tivoli Storage Manager, very quick to restore data from the array when users needed files restoring and also the added bonus of robust off site storage of encrypted company data.

I'm guessing the LTO5 solutions came over budget? How much is the data worth to the company out if interest?
 
I recently went from LTO2 to pure Disk to Disk. We have the advantage of 2 separate locations on a large site (factory and office).

Each site has a server room with aircon and fire protection. Each backup runs either every 8 hours (production databases) or daily. All backups are pulled by Backup Exec to the secondary backup server with 7TB of disks. We have 2 ESXi hosts in the backup location too which hold backup copies of each physical server so in the event of system loss they can be booted and backups restored. Lower performance but at least they will be running! Worst case scenario is I lose up to 8 hours of production data and 24 hours of other data. But advantage is I can restore pretty much everything in under 4 hours.

Even with the above I am considering an LTO4 for the critical stuff to run a duplicate job onto in order to protect the data. Option for next year is a SAN in each location running a network mirror and a full virtualisation project to move into vSphere 5.0 so we would effectively just failover with no data loss.

If we lost both locations then the data does not matter as it would take us at least a year to rebuild anyway. All financials are stored in our ERP on remote servers at our datacentre so we are covered for audits.

Cost should not really be a factor unless you a talking 10's of thousands. How much would it cost your business if you lost it all?
 
Last edited:
Jay794 said:
Theft? hardly, the servers are the size of fridges and the facilty has security on site 24/7. Fire? possibly except the server room has one of those vacuum systems to remove all the oxygen from the room, thus stopping the fire. Flood? possibly, I suppose if it rained hard enough the upstairs could flood but the roof would have to cave in first
Click to expand...

Although you appear to have reasonable protection it is vital that these factors are considered. I know of 2 companies that lost production and data due to flooding recently; one because they located the Comms Room under toilet facilities (whoops :) and another due to a leak on the air con cooling the room. The fact is that it can happen to anyone and IT staff should always store key data off-site. I'm also dubious of fire proof safe's. Can they really protect flimsy plastic in an inferno? Maybe I'm a cynic or maybe just too old, but I can't see it.
 
aspirin said:
and another due to a leak on the air con cooling the room.
Click to expand...
Years ago on one of the projects he ran my Dad had a load of servers that were being configured for the customer and they were put at the far end of an open-plan floor away from everyone (because of the noise) for configuring. The aircon units above them decided to dump all their coolant in the servers, but they only noticed because of the remote monitoring screens all lighting up red. Amazingly, thanks to having redundant everything in them, the servers kept running!



aspirin said:
I'm also dubious of fire proof safe's. Can they really protect flimsy plastic in an inferno? Maybe I'm a cynic or maybe just too old, but I can't see it.
Click to expand...
Same - I don't really get fireproof safes. They'll keep the actual fire away from your tapes, but they won't be able to keep it cool...
 
Tapes are tried and tested. Been around for years, and a LTOX tape will last for years.

I would only use disk for temporary staging areas before the data gets duplicated to tape. I was looking into having a dedupe solution where my 6 week cycle gets deduped to disk using a Symantec NBU device, and then once a month we'd create a synthetic full out of all the deduped data, cutting down on the cost of tapes drastically.
 
I use Backup Exec to Tape.

Iron Mountain then collect my 5 (4 weekly & 1 monthly) tape at the start of every month. throughout the month, the tapes are stored securely in a fireproof safe. The Iron Mountain man collects the tapes and places them in a robust briefcase type thing then stores them offsite.

Tapes are more robust than HDD if you ask me. I'd stick with tapes.
 
You must log in or register to reply here.
Back
Top Bottom