Slashdot Mirror
Build Your Own 135TB RAID6 Storage Pod For $7,384
An anonymous reader writes "Backblaze, the cloud-based backup provider, has revealed how it continues to undercut its competitors: by building its own 135TB Storage Pods which cost just $7,384 in parts. Backblaze has provided almost all of the information that you need to make your own Storage Pod, including 45 3TB hard drives, three PCIe SATA II cards, and nine backplane multipliers, but without Backblaze's proprietary management software you'll probably have to use FreeNAS, or cobble together your own software solution... A couple of years ago they showed how to make their first-generation, 67TB Storage Pods"
for both internet security and privacy: each of us can now store his own local copy of the internet and surf offline!
The article says it uses RAID 6 - 45 hard drives are in the pod, which are grouped into an arrays of 15 that use RAID 6 (the groups being combined by logical volumes), which gives you an actual data capacity of 39TB per group (3TB * (15 - 2) = 39TB), which then becomes 117TB usable space (39TB * 3 = 117TB). The 135TB figure is what it would be if you used RAID 1, or just used them as normal drives (45 * 3TB = 135TB).
And these are all "manufacturer's terabytes", which is probably 1,024,000,000,000 bytes per terabyte instead of 1,099,511,627,776 (2^40) bytes per terabyte like it should be. So it's a mere 108 terabytes, assuming you use the standard power-of-two terabyte ("tebibyte', if you prefer that stupid-sounding term).
This is nothing new. You've never been in a datacenter before, kid. You can ask a grownup one day and he can take you there and you will feel the heat. And NOISE. No offense, but I think you're one of those gamer kids who builds rigs for max FPS, with esoteric water cooling and silent fans everywhere.
Yeah, no, you don't need to pamper your hardware that much. Even laptop drives work way hot (60C+) for years with no issue.
Most servers are built that way too. The Sun x4500 is extremely densely packed. And there are hundreds running just fine.
If you're in the SF Bay Area check out http://geeksessions.com/ where Gleb Budman from Backblaze will be speaking about the Storage Pod and their approach to Network & Infrastructure scalability along with engineers from Zynga, Yahoo!, and Boundary. This event will also have a live stream on geeksessions.com.
Full Disclosure: This is my event.
50% discount to the event (about $8 bucks and free beer) for the Slashdot crowd here: http://gs22.eventbrite.com/?discount=slashdot
Here is a link to Backblaze's actual blog entry for the new pods 135TB, and here is the original 67TB pods. The blog article is actually quite fascinating. Apparently they are employee owned, use entirely off-the-shelf parts (except for the case, looks like), and recommend Hitachi drives (Deskstar 5K3000 HDS5C3030ALA630) as having the lowest failure rate of any manufacturer (less than 1% they say).
I found it kinda amusing that ext4's 16TB volume limit was an "issue" for them. Not because its surprising, but because... well, its 16TB. The whole blog post is actually recommended reading for anyone looking to build their own data pods like this. It really does a good job showing their personal experience in the field and problems/not problems they have. For instance: apparently heat isn't an issue, as 2 fans are able to keep an entire pod within the recommended temperature (although they actually use 6). It'll be interesting to see what happens as some of their pods get older, as I suspect that their failure rate will get pretty high fairly soon (their oldest drives are currently 4 years old, I expect when they hit 5-6 years failures will start becoming much more common.) All in all, pretty cool. Oh, and it shows how much Amazon/ Dell price gouges, but that shouldn't really shock anyone. Except the amount. A petabyte for three years is $94,000 with Backblaze, and $2,466,000 with Amazon.
P.S. I suspect they use ext4 over ZFS because ZFS, despite the built in data checks, isn't mature enough for them yet. They mention they used to use JFS before switching to ext4, so I suspect they have done some pretty extensive checking on this.
"None can love freedom heartily, but good men; the rest love not freedom, but license." --John Milton
Hardware RAID controllers are stupid in this context. The only place they make sense is in a workstation, where you want your CPU for doing work, and if the controller dies you restore from backups or just reinstall. Using software RAID means never having to try to get a rebuilder software to convert the RAID from one format to another because the old controller isn't available any more, or because you can't get one when you really need one to get that project data out so you can ship and bill.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Because, for this project, raw storage capacity is much more important than performance. Besides, they claim their main bottleneck is the gigabit Ethernet interface - even software RAID, the PCIe x1, and the raw drive performance is less of a limiting factor.
Yeah, in a situation where you need high I/O performance, this design would be less than ideal. But they don't - they're providing backup storage. They don't need heavy write performance, they don't need heavy read performance. They just need to put a lot of data on a disk and not break anything.
PS: SAS doesn't really provide much better performance than SATA, and it's a lot more expensive. Same for hardware RAID - using those would easily octuple the cost of the entire system.
My problem with Backblaze is their marketing is very misleading...they pit these storage pods up against cloud storage and assert that they are "cheaper", as though a storage pod is anything like cloud storage. It isn't. Sure, there's the management software issue that's already been mentioned, but they do no analysis on redundancy, power usage, security, bandwidth usage, cooling, drive replacement due to failure, administrative costs, etc. It's insulting to anyone who can tell the difference, but there are suits out there who read their marketing pitch and decide that current cloud storage providers like Google and Amazon are a rip off because "Backblaze can do the same thing for a twentieth the price!" It's nuts.
You can see this yourself in their pricing chart at the bottom of their blog post. They assert that Backblaze can store a petabyte for three years for either $56k or $94k (if you include "space and power"). And then they compare that to S3 costing roughly $2.5 million. In their old graphs, they left out the "space and power" part, and I'm sure people complained about the inaccuracies. But they're making the same mistake again this time: they're implicitly assuming the cost of replicating, say, S3, is dominated by the cost of the initial hardware. It isn't. They still haven't included the cost of geographically distributing the data across data centers, the cost of drive replacement to account for drive failure over 3 years, the cost of the bandwidth to access that data, and it is totally unclear if their cost for "power" includes cooling. And what about maintaining the data center's security? Is that included in "space"?
On a side note, I'd be interested to see their analysis on mean time between data loss using their system as it is priced in their post.
You could say the Backblaze is serving a different need, so it doesn't need to incur all those additional costs, and you might be right, but then why are they comparing it to S3 in the first place? It's just marketing fluff, and it is in an article people are lauding for its technical accuracy. Meh.
Or can somebody tell me if the cooling of the HDs is ok if they are stacked like in the picture?
According to their blog post about it, they see a variation of ~5 degrees within unit (middle drives to outside drives) and about 2 degrees from the lowest unit in a rack to the highest. They also indicate that the drives stay within the spec operating temperature range with only two of the six fans in each chassis running.
Keep in mind these are 5400 RPM drives, not the 10K+ drives you would expect in an application where performance is critical. These are designed for one thing - lots of storage, cheap. No real worries about access times, IOPS, or a lot of the other performance measures that a more flexible storage solution would need to be concerned with. These are for backup only - nice large chunks of data written and (hopefully) never looked at again.
Did you notice how they even gave you the alternatives to their software? Essentially they are saying "We developed this for our own internal use and if you would LIKE to pay for it its cool. If you dont then there are these other free alternatives." But then again just because some company is mentioned in the article it MUST be a slashvertisment.
No. Hardware controllers are the right solution in this context. These pods are not designed for individual users, but for corporations that can afford stockpiles of spare parts, so replacing a board can be done easily. Using hardware controllers allows many more drives per box, and thus per CPU. A populated 6-CPU motherboard is going to be less reliable, dissipate more heat, require more memory, and likely be less reliable, than the special-purpose hardware approach that allows for a single CPU.
Software RAID makes sense when you have a balance of storage bandwidth requirements to CPU capacity that is heavy on the CPU side. This box is designed for the opposite scenario, as the highly informative blog describes:
http://blog.backblaze.com/2011/07/20/petabytes-on-a-budget-v2-0revealing-more-secrets/
(Yes, I know, expecting someone to read the blog would mean that they would have to read the linked article and then click through to the original post, a veritable impossibility. Still, it is recommended reading, especially the part about their experience with failure rates and how they have *one* guy replacing failed drives *one* day per week.)
Put my fist through my alarm clock with its ding-dong death inside my ear. - The Blackjacks.
First: ALL Marketing is misleading. That is what marketing does. Accentuate the positive, eliminate the negative. So complaining about that is just idiotic.
Second: You could have a couple dozen Backblaze units, pay for a tech to monitor them 24/7/365 and replace all the drives twice over for what Amazon charges for the same thing. Sure that doesn't included cost for premises, and HighSpeed Internet to multiple locations. But still, that is aggregated with all the other clients.
Third: what are you paying for in the "cloud", I mean besides ethereal concepts. Does Amazon tell you how they do things? You probably know less about Amazon (and the others) setup so you're comparing something you know something about (not everything) verses something you know almost nothing about, and the complain that they aren't doing it in a comparable way. You don't know.
Fourth: Your basic assumption is that Backblaze has no contigency for drive replacement, which is false. Since these are "new" drives there might be insufficient data about failure rates and therefore the actual cost of replacement (never mind warranties) or having drives in both Hot and Cold Spare setups. I'm sure that Backblaze in their $5/MO service figures what it costs to store data, have spares, keep the Datacenter running and profitable. Even if they double the cost to $10, it still puts the others to shame.
Have you compared the data loss rates for the last three years between Amazon and Backblaze? Can you even compare or is that data held secret (see point 1b). My point here, is that you're pulling shit out of your ass and thinking it doesn't stink. Even if it isn't directly comparable, it is at least in the realm of consideration, EVEN if everything you said is true. And at 10 times less in cost, that can buy a lot of redundancy. It is just a matter of perspective.
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
... if you really care about the data.
(Disclaimer: I work at Backblaze) - If you really care about data, you *MUST* have end-to-end application level data integrity checks (it isn't just the hard drives that lose data!).
Let's make this perfectly clear: Backblaze checksums EVERYTHING on an end-to-end basis (mostly we use SHA-1). This is so important I cannot stress this highly enough, each and every file and portion of file we store has our own checksum on the end, and we use this all over the place. For example, we pass over the data every week or so reading it, recalculating the checksums, and if a single bit has been thrown we heal it up either from our own copies of the data or ask the client to re-transmit that file or part of that file.
At the large amount of data we store, our checksums catch errors at EVERY level - RAM, hard drive, network transmission, everywhere. My guess is that consumers just do not notice when a single bit in one of their JPEG photos has been flipped -> one pixel gets every so slightly more red or something. Only one photo changes out of their collection of thousands. But at our crazy numbers of files stored we see it (and fix it) daily.