How to connect LTO drive to server

Topic

New Reply

Can someone tell me how this backup device is supposed to connect to a server? I’ve never setup one of these and an associate of mine is looking to buy one. It’s unclear to me what he’ll need to connect it to the server. He has a RAID controller in his server (a small HP tower server) and it has an external connection to it, but I don’t think you’re supposed to connect this to a RAID controller. Maybe I’m wrong. I think he’ll need to buy an add-on card, but what kind? That’s where I’m stuck.

Tandberg
LTO 3 HH External

Tandberg
LTO 4 HH External

Reply | Quote

Replies

Tandberg LTP tape drives usually have either a Fibre Channel, SAS or SCSI interface.

Reply | Quote

You will most likely need an external SCSI interface card. The most common is LVD (Low Voltage Differential). You need a SCSI terminator for these.
If you can tell us the tape model we can advise.

cheers, Paul

Reply | Quote

Here’s the make/model of the one from Newegg he’s looking at. It’s a SAS connected device so I guess he just purchases a SAS add-on card to go with it?

http://www.newegg.com/Product/Product.aspx?Item=N82E16840119042

Reply | Quote

Yes, you need an external SAS interface card and an external SAS cable.
I would also suggest backup software rather than the Windows product.

cheers, Paul

Reply | Quote

According to

[indent] Tandberg’s SAS HBA Compatibility Matrix[/url][/indent]

you need either a

[indent]LSI LSI00188 PCI Express Low Profile Ready SATA / SAS 9200-8e Controller Card (Single Pack) – $334.99[/indent]

or a

[indent]ATTO Technology ExpressSAS 6 Gbps SAS/SATA PCIe 2.0 Host Bus Adapter (8 External Ports) – $319.00[/indent]

Refer to your

[indent]Tandberg Data LTO-3, LTO-4, & LTO-5 HH SAS External Tape Drives Quick Start Guide[/indent]

that shipped with your Tandberg 3513-LTO for connection instructions.

More info can be found on

[indent]this Tandberg page[/indent]

(see the “Documents” link on the right).

Reply | Quote

Thanks for all the help everyone. One last question…

I’m sure there’s a big speed difference between an LTO 3 or 4 tape backup system and a RDX USB 3.0 backup system. However, there seems to be about a $1,000 price difference when you add in everything (backup hardware, cartridges or tapes, and add-on cards). For a small business with 1 server and 10 users do you feel like a RDX USB 3.0 backup system would be at least an okay idea if not a good idea?

Reply | Quote

LTO vs RDX

In General, RDX is most suitable for clients who want to backup smaller datasets and don’t need to archive lots of data for an extended period. LTO is a better option for clients with large datasets to backup, and / or who want to archive data for extended periods. Basically for a smaller data set, say 100GB daily backup / 400GB weekly with 6 tapes / cartridges in rotation the RDX solution will provide a lower cost solution. However let’s say you’re backing up 400GB daily / 1.3TB weekly with 12 tapes / cartridges, then the LTO Tape Solution is going to provide a lower cost solution.

Reply | Quote

Personally I’d be using external hard disks to backup the data. This will be much cheaper and easier than a tape unit – tapes alone are >$50 a pop and that will buy you a large USB hard disk. You can also train users to swap hard disks on their local PC and run the backup overnight.
Tape units and software are a lot of work for a small organisation.

cheers, Paul

Reply | Quote

For long term storage of your data (i.e. years), tape is a great way to go. The big advantage is that with tape, you are storing only the medium which holds the data. With hard drives, you are storing the medium plus the drive. The drive could go bad on the shelf, and then you’d have a heck of a time trying to get to your data. There’s a lot less likelihood that a tape will go bad. And if you clone your tapes, you’ll have two copies, virtually assuring that you’ll have a good copy of the data.

According to Curtis Preston, tape is a lot cheaper than hard drives:
http://backupcentral.com/mr-backup-blog-mainmenu-47/13-mr-backup-blog/412-get-rid-of-tape-inconceivable.html

Group "L" (Linux Mint)
with Windows 10 running in a remote session on my file server

Reply | Quote

Rats – getting behind again on discussions here.

For long term storage of your data (i.e. years), tape is a great way to go.[/quote]

But for relatively modest applications such as the one under discussion here disk is probably far better (that can be true for larger installations as well, for that matter).

The big advantage is that with tape, you are storing only the medium which holds the data. With hard drives, you are storing the medium plus the drive. The drive could go bad on the shelf, and then you’d have a heck of a time trying to get to your data. There’s a lot less likelihood that a tape will go bad.

Perhaps, but people serious about their data won’t be depending upon a single copy anyway and will be periodically ‘scrubbing’ that data to detect and correct bit-rot while the other copy or copies are still available to fix it.

And if you clone your tapes, you’ll have two copies, virtually assuring that you’ll have a good copy of the data.

Ditto for disks, but with disks it’s far easier to use parity-based RAID-6 to create three effective copies with little more space overhead than that required to store a single copy – or for the truly paranoid dual independent RAID-5s (or RAID-6s) to create four (or six) effective copies in two completely independent hardware environments (updated independently) with little more space overhead than that required for two simple copies.

Add in the fact that with suitable management such disk storage can satisfy both backup and archival needs plus additional facilities like continuous data protection and deduplication which tape just isn’t very good at and the choice of disk in situations where multiple such needs exist and do not intrinsically benefit from separate solutions is pretty much a no-brainer.

According to Curtis Preston, tape is a lot cheaper than hard drives:
http://backupcentral.com/mr-backup-blog-mainmenu-47/13-mr-backup-blog/412-get-rid-of-tape-inconceivable.html

If Curtis Preston has had an original idea in the past couple of decades I must have have missed it (though admittedly he might have, since I tend to ignore him unless I encounter him in some other context). But that’s OK – repeating industry information has its place: it’s his frequent abject incompetence that actively annoys me (in marked contrast to his EMC colleague Stephen Manley, whose own blog post he links to). The only explanation I can find for his continued apparent popularity is that his audience must consist of management types who are as technologically clueless as he is:

Just to compare cost, at $35 per 1.5 TB tape, storing 20 PB on LTO-5 tapes costs $22K with no compression, or $11K with 2:1 compression.

Anyone who completed grammar school should have better arithmetic skills than this: at the quoted $35/1.5TB 20PB should cost over $466K (without compression), not $22K. Being off by well over a decimal order of magnitude is a bit beyond my normal tolerance for sloppiness.

And note that that’s ignoring the cost of the tape drive itself – but when pricing disk storage (at $625/TB) he didn’t forget to include the full cost of the associated infrastructure rather than simply the cost of the storage units themselves (which at $30 – $40/TB for vanilla-flavored desktop 3.5″ drives – and less than twice that for ‘enterprise’ SATA storage, though with suitable redundancy that’s arguably overkill – is not all that much higher than his above quote for tape).

Making another copy on tape at $.013/GB (current LTO-5 pricing)

Wait a minute: didn’t he just quote us $35/1.5TB above. That’s more like $.0233/GB, but at least it’s only off by a factor approaching 2 this time rather than more than ten times that.

The disk array is much more expensive than a tape

Only if you buy it from someone like EMC. By contrast, Linux can create a robust array from individual drives for only the cost of the drives themselves plus their cables – i.e., at a cost that’s VERY competitive with tape once you include the cost of the tape drive.

While I wouldn’t suggest sending your disks to Iron Mountain unless they can guarantee VERY careful handling throughout, if their price for storing tapes is really $8/TB/year a small business can quite cost-effectively replicate its data on disks and gently ferry them to a couple of external locations where (using no more power than stored tapes, unless you’d like them to remain online) they’ll be FAR more readily accessible (and then immediately usable) should it need them on short notice.

If I put it on disk, I’m going to need to buy a new disk every five years and copy it.

About what I’d expect from Curtis. A disk’s nominal service life of 5 years assumes that it’s spinning for all (or at least a good percentage) of that time: if the disk is spun up for a few hours every few months for ‘scrubbing’ purposes its effective life will be far longer. Not, of course, that it matters all that much, because after 5 years its capacity will be replaceable for only a small fraction of its original cost and the increase in the amount of the data needing to be stored will likely make using newer, significantly larger disks at far lower cost/TB the right choice even if the old disks would last forever.

By contrast, I wouldn’t expect the cost/TB of tape storage to decrease at anything like the same rate, but since that’s not an area I’m as familiar with I’d be happy to see evidence to the contrary.

I would also argue that if one error every 10 PB is too much, then you can make two copies — at a cost an order of magnitude less than doing it on disk.

Leaving aside his continuing inability to compare tape apples (cost of a tape) to disk apples (cost of a disk), only Curtis could think that anyone at all serious about their data would be satisfied with only a single additional copy of it. And once you DO have a second copy you get MANY decimal orders of magnitude more security against unrecoverable bit errors, completely eclipsing the mere 2 decimal orders of magnitude difference he claims between the nominal bit-error-rates of disk and tape.

disks use significantly smaller magnetic grains than tape, and disks run at very high operating temperatures, where tape is stored in ambient temperatures.

Yet more evidence that arithmetic is not Curtis’s strong suit, I’m afraid. While the smaller grain size can make a noticeable difference (though I expect that the ability of disks to orient their grains vertically in higher-coercivity material – which the thinness of tapes presumably makes much more difficult to achieve – compensates to a significant degree for this), the difference between ‘ambient’ temperature (say, 20C) and the ‘very high’ disk operating temperature (25C – 35C) is negligible, because the temperature in question is measured not from 0C but from 0K (absolute zero, or -273C) and thus the difference is only a few percent. A sterling exemplar of the observation (in many slightly different phrasings) that “a little knowledge is a dangerous thing”.

The result is that disk cannot be trusted to hold onto data for more than five years without suffering bit rot.

Anyone who thinks that disks CAN be trusted to hold data for 5 years without suffering from bit-rot doesn’t know what he’s talking about (but we already knew that…). It’s a matter of probabilities, and while there’s a good chance that no bit-rot will occur within 5 years I certainly wouldn’t count on it if I were serious about my data.

That’s where ‘scrubbing’ comes in, because by reading the entire disk every few months the disk’s firmware has the opportunity to detect areas which are slowly becoming difficult to read and rewrite them (refreshing the magnetic grains) before they become impossible to read (a process which is more difficult to perform with tape: good old DECtapes could have their sectors randomly re-written but I’m not sure that conventional tapes can – in which case a tape with a single rotten bit would have to be rewritten in its entirety). With scrubbing, disks can avoid bit-rot as well as (perhaps better than) unscrubbed tape can.

Only optical (i.e. non-magnetic) formats (e.g. BluRay, UDO) do a better job of holding onto data for decades. Unfortunately they’re really expensive. Last I checked, UDO media was 75 times more expensive than tape.

Can’t speak for UDO, but write-once BluRay storage (like write-once DVD storage – in both cases more stable than rewritable versions) is (and has been for quite a while) often available at under $0.04/GB – looks as if Curtis is once again off by more than a decimal order of magnitude even if one doubles that to try to ensure dyes with the best longevity (though last I know how important this is is a subject of continuing debate).

As for his observations about an ‘air gap’ (not that you couldn’t do the same with disk, of course), does he seriously think that ANY form of backup can protect a system against a malicious system admin?

Apologies for having torn what passes for his expertise apart at such length: as I said, his incompetence (and the degree to which it seems nonetheless to be popular) annoys me, perhaps especially because it’s so emblematic of what passes for expertise in so many other (and often far more important) areas of our society these days.

Reply | Quote