Storage & Backup Ideas

[BlueFireAngel]

I work for a small construction company that is looking to upgrade their storage and backup processes and systems. I'm looking for some options.
Background:
We are a fairly small company; about 60 employees in the field, 10 in the office. We have as many as 6-10 mobile workstations that are primarily for electronic record reporting. Of those, two can be used as supplemental estimating workstations if necessary. In the office, there are 3 workstations with estimating as their primary function. These access the file server regularly because all project information is primarily stored on that server. The other 7 workstations in the office are project administration or support machines. Only two workstations run Win7 OS; one laptop and one desktop. The servers run Server 2003.
The file server is the primary storage repository for all company and project information. Each workstation has a hard drive of approximately 100-500GB for various applications ranging from estimating software to CAD software to Office software. The server has ~1TB HDD. It is likely that all office workstations and the server will be upgraded in order to be compliant with Win7 and Server 08 respectively, within 2 years. The company is small enough that this could be done all at once or in two groups depending on funds.
We currently use 200GB tapes to backup the server nightly. There are 6 tapes, one each night monday through friday, and one for the weekend that are stored in a fire/water proof safe on the premises. Tapes are switched out each morning. The workstations are only backed up by manually sending files to the server.
Looking Forward:
As the construction industry continues into the digital age, we are storing more and more information digitally. This includes CADD design and modeling files, jobsite photos & video, inspections by video, company and municipal forms, financial information, contracts & subcontracts, invoices, etc. All of this information needs to be stored and backed up digitally and securely. We'll soon reach our tape limit and the question becomes, then what? What option will lend itself to our projected growth.
Options:
I'm not an IT guy and I don't have the most experience with servers and networking systems. I am trying to expand my skill set and I plan to use similar strategies at home for my personal files and multimedia. This company was a small family business. It's on the edge of expanding beyond that and so they are beginning to expand and upgrade their infrastructure so that they can accommodate future growth.
Here are the options I see so far:
--Use 1-2 TB external HD to backup the server. Use 1TB external HD to backup each workstation or 1 external HD for several workstations to share by manually conntecting or NAS
--Larger tapes or span a backup across multiple tapes.
--Use a RAID 1 or greater setup for the next server.
--Use online client such as Mozy or Carbonite.

What backup strategies do you use? Software, hardware. Onsite vs Offsite, etc.?
What other options are there for company scale data storage and backup?

I'll be happy to try to answer any questions. Thanks for the help.

 

[sminlal]

For small businesses, external hard drives are a pretty compelling solution for backup media. They're cheap, have a lot of capacity, and in a pinch you don't need to find a machine with just the right kind of tape drive in order to restore your data. Yes, they can fail, but so can tapes. In my experience tapes become problematic because not only does the tape wear out but it also wears out the heads in the tape drive too - those problems don't exist with hard drives. Hard drives are somewhat more delicate than tapes, but when they're spun down they're more robust than most people give them credit for.

The key with any media is to rotate your backups through multiple media sets and keep multiple cycles of backups so that if one set fails you have another to fall back to. And if you keep backup rotations for daily, weekly and monthly cycles then you can go back and recover files which were accidentally deleted or were improperly edited (or whatever) as far back as you keep cycles for.

It's important not to treat RAID 1 as a backup. RAID 1 protects ONLY against disk failure, not against a host of other risks such as accidental deletion, corruption, theft, or common-mode failures such as power hits that fry the system and its drives. To protect against all those risks you need to back up to media which is then stored offline, preferably with at least one copy stored offsite. Offsite backup doesn't necessarily have to be fancy, it can literally be as easy as the IT guy taking a copy home with him (but watch out for privacy issues).

If you do that, then you don't really need RAID 1 to protect your data, what you need it for is to protect your UPTIME. If your business looses money when your central server is down, RAID 1 is part of your insurance against that.

I personally don't trust Internet-based backup services. If you look at the license agreements I suspect you won't be able to find one that actually gives you any sort of solid guarantee of data availability. If you use them, treat it as a "Plan B" and not as your only backup methodology.

 

[BlueFireAngel]

The key with any media is to rotate your backups through multiple media sets and keep multiple cycles of backups so that if one set fails you have another to fall back to.

I'm not sure what you mean by this. Do you mean backup on separate HD's (if that's the way we choose to go) on separate days or weeks? For instance, if we kept doing the nightly backup routine with external HD's instead of tapes, we would need 6 external HD's.

I definitely agree with you on the internet based backup services. They kinda freak me out as a primary backup method. I think I would only seriously consider it for a small number of important documents that I had in several other places and that I didn't need to update often.

Thanks for the reply.

 

[sminlal]

Yes, I mean multiple hard drives. If one hard drive fails, you go to the previous backup on a different hard drive, the same way you would if a tape failed on you.

But because hard drives are random access, you don't necessarily need 6 hard drives for daily backups in a weekly cycle. You could, for example, use just two hard drives and alternate between them every other day (assuming they have the capacity to hold 3 daily backups each).

Of course the more drives you use, the better protected you are and the more previous backup versions you can keep - those can help you to recover from problems that you don't discover right away. So you have to make a decision about what kind of cost vs. data security you're comfortable with.

In the grand scheme of things hard drives are pretty cheap - you can buy 2TB hard drives for about $100 now.

Pay attention to the data integrity specs of the hard drives. A lot of consumer class hard drives are rated at 1 unrecoverable read error per 10^14 bits read, which means on a 2TB drive you'd expect to get one read error when reading about 5 complete drives' worth of data. Others are rated at 1 per 10^15 bits. I rather like the WD Green drives because they have the latter rating, are cheap, and run cool (something to consider for drives in external enclosures).

 

[kai-bcn]

Hi,
I work on the CA ARCserve recovery management family, and highly recommend thinking through all the different events & incidents which would prompt you to restore data, and then designing a solution that would address them ALL.

When thinking about RAID levels and hard drive error rates, that's just one category - the one where individual IT components are failing, and how you recover from the situation. If data loss is involved, how quickly can you restore, and what version / what "restore point" do you have after the restore? A typical backup that runs daily will achieve Recovery Points of up to 24 hours. Real-Time data protection / Replication / HA / CDP can achieve better results, if the business things that losing data of up to 1 full business day is not acceptable. Into this first category the hardware purchasing & vendor response times also play a role. If a server is broken, even though the disk inside is fine, how quickly can it get repaired to have it running again? Or is it quicker to restore to another one? Or can High Availability simply switch over to a stand-by server that has the latest copy of modifications already?

Category 2 of backup needs could be classified as the need for historic data. That's why you currently write daily tapes, and rotate them out. If done nicely (GFS etc) you will have additional versions (to the on above) to restore from, such as last week, last month, and older. The same can be done with tapes.

Category 3 might be a full site disaster - either natural or manmade. How would you recover? That's the one that most easily gets overlooked when moving from tapes to disk-based solutions. Either data gets replicated between different sites, or has to move physically on a removable media.

One way for you to continue using the tapes even though the capacity limit is hit, is this: with backup programs such as ARCserve Backup you can specify a first copy to go to disk, including data deduplication, so that various versions / recovery points can be hold there - for example up to 2-4 weeks. Those would then automatically be used for restores. Instead of writing 1 tape each day, you can then specify to only copy data from disk to tape once a week. If the backup doesn't fit a single tape, then 2 tapes or more will be used (known as tape spanning). These can then be moved to a location that is geographically different to the servers, in order to be protected for cases such as employees going mad, theft, fire, flooding, tornados or terrorism.

The quickest full server restore to completely new hardware or even from Physical to Virtual machines can be achieved with ARCserve D2D, an image level restore.

If even the quickest restore time is too long to wait for your business users, then Replication & High Availability will provide a great level of protection. Rather than just protecting one disk with another disk (RAID), you protect one server with another server - and hence even in a mainboard / CPU issue event, you have another system to provide services with.

Hope this helps, Kind Regards, Kai