What is Digg?Sponsored by Best Buy
Think Choosing a Gift For A Teen Girl Is Impossible? view!
bestbuy.com - Hello-o-o! No WAY! Email, IM, chat, social networking & streaming audio/video are all on Netbooks.
231 Comments
- diggB, on 09/04/2009, -39/+699………………….._,,-~’’’¯¯¯’’~-,,
………………..,-‘’ ; ; ;_,,---,,_ ; ;’’-,…………………………….._,,,---,,_
……………….,’ ; ; ;,-‘ , , , , , ‘-, ; ;’-,,,,---~~’’’’’’~--,,,_…..,,-~’’ ; ; ; ;__;’-,
……………….| ; ; ;,’ , , , _,,-~’’ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ¯’’~’-,,_ ,,-~’’ , , ‘, ;’,
……………….’, ; ; ‘-, ,-~’’ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;’’-, , , , , ,’ ; |
…………………’, ; ;,’’ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;’-, , ,-‘ ;,-‘
………………….,’-‘ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;’’-‘ ;,,-‘
………………..,’ ; ; ; ; ; ; ; ; ; ; ; ;__ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ‘-,’
………………,-‘ ; ; ; ; ; ; ; ; ; ;,-‘’¯: : ’’-, ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; _ ; ; ; ; ;’,
……………..,’ ; ; ; ; ; ; ; ; ; ; ;| : : : : : :| ; ; ; ; ; ; ; ; ; ; ; ; ,-‘’¯: ¯’’-, ; ; ;’,
…………….,’ ; ; ; ; ; ; ; ; ; ; ; ‘-,_: : _,-‘ ; ; ; ; ; ; ; ; ; ; ; ; | : : : : : :| ; ; ; |
……………,’ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ¯¯ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;’-,,_ : :,-‘ ; ; ; ;|
…………..,-‘ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ,,-~’’ , , , , ,,,-~~-, , , , _ ; ; ;¯¯ ; ; ; ; ;|
..…………,-‘ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;,’ , , , , , , ,( : : : : :) , , , ,’’-, ; ; ; ; ; ; ; ;|
……….,-‘ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;’, , , , , , , , ,’~---~’’ , , , , , ,’ ; ; ; ; ; ; ; ;’,
…….,-‘’ ; _, ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ‘’~-,,,,--~~’’’¯’’’~-,,_ , ,_,-‘ ; ; ; ; ; ; ; ; ; ‘,
….,-‘’-~’’,-‘ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; | ; ; | . . . . . . ,’; ,’’¯ ; ; BUT CAN THEY HANDLE A PETAFILE?
……….,’ ; ;,-, ; ;, ; ; ;, ; ; ; ; ; ; ; ; ; ; ‘, ; ;’, . . . . .,’ ;,’ ; ; ; ;, ; ; ;,’-, ; ;,’ ‘’~--‘’’
………,’-~’ ,-‘-~’’ ‘, ,-‘ ‘, ,,- ; ; ; ; ; ; ; ; ‘, ; ; ‘~-,,,-‘’ ; ,’ ; ; ; ; ‘, ;,-‘’ ; ‘, ,-‘,
……….,-‘’ ; ; ; ; ; ‘’ ; ; ;’’ ; ; ; ; ; ; ; ; ; ; ‘’-,,_ ; ; ; _,-‘ ; ; ; ; ; ;’-‘’ ; ; ; ‘’ ; ;’-,
……..,-‘ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;¯¯’’¯ ; ; ; ; ; ; ; ; , ; ; ; ; ; ; ; ; ;’’-,
……,-‘ ; ; ; ; ; ; ; ,, ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; |, ; ; ; ; ; ; ; ; ; ; ‘-,
…..,’ ; ; ; ; ; ; ; ;,’ ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;|..’-,_ ; ; ; , ; ; ; ; ; ‘, - techdever, on 09/04/2009, -5/+248Finally, some proper tech news on digg...
Getting tired of the gay acid trip race and cats - AlyxVance, on 09/04/2009, -6/+181At first I was like "wtf is pedobear doing in a tech thread"
and then was like :D - SpoonMSU, on 09/04/2009, -2/+170OVER 9000
- inactive, on 09/04/2009, -3/+124BRILLIANT!!!
- NiftyG, on 09/04/2009, -1/+109Their costs seem to be at least 80% less per petabyte than other major manufacturers. Maybe this company should start a side business selling servers or server kits. Those rack enclosures are pretty sweet.
- antivirus88, on 09/04/2009, -1/+108This was an AWESOME article. I want to see more stuff like this on Digg.
- zimsters, on 09/04/2009, -9/+115That's a lot of porn.
- swizzcheez, on 09/04/2009, -2/+101You must be joking about me joking:
http://hardware.slashdot.org/story/09/09/02/138209 ...
This was on /. on Wednesday, it doesn't work its way up through digg until Friday...
The whole reason I started reading digg was because it was faster than slashdot to get news out. Now it seems the only news that makes it to the front page quickly is politics (esp. left leaning) and humor links.
Just sayin'... - MrDoug, on 09/04/2009, -16/+114Builded....really.....you builded?
- swizzcheez, on 09/04/2009, -6/+86It used to be slashdot lagged digg. Now it's the other way around for anything actually technical... :/
- PowderedToasty, on 09/04/2009, -1/+60fejorca: A male from Colombia
- Flytrap, on 09/04/2009, -0/+58That is the wonderful thing about the English language, it lends itself to being understood easily even when English is not the speaker's (or writer's, in this case) first language.
We all know what he meant to say... it is clear that he got his tenses wrong, but the content and intent of his message still came through - at least to those of us for whom English is a first language.
If you have kids you will know that the tense error is not a function of intelligence, it is a function of exposure and practice... my 2 year old has "builded" houses for the "fishes" and "sheeps" to live in many times. - tommythetomcat, on 09/04/2009, -3/+60thats ironic, because i have you blocked
- trafficlight, on 09/04/2009, -3/+58Well you deserve a cookie.
- inactive, on 09/04/2009, -0/+50In 10 years, this will seem like a 3.5" floppy disk.
- candre23, on 09/04/2009, -2/+43They've had a couple petabytes online for the last few years without any problems. Clearly they know something you don't.
- groo68, on 09/04/2009, -0/+39Except old articles of course.
- thundercloud39, on 09/04/2009, -3/+41I think this proves beyond a shadow of doubt that pedobear can always be applied to any article.
- redux2redux, on 09/04/2009, -0/+38A refreshing change from let's try and patent /sue to share the design.
- fejorca, on 09/02/2009, -7/+44I builded storage servers for a video production company, is amazing that I used almost the same components and the only difference are I use intel DG33FB motherboards and FreeNAS x64, nice job!
- SirBruce, on 09/04/2009, -3/+39As a former NetApp employee, I want to take a minute to defend the high cost of professional storage servers. What you're paying for if you buy such a server from NetApp or EMC is performance and reliability more than the "actual space".
Yes, the Backblaze system is cheap. It will also suffer from many problems:
1. Performance will be terrible. Mind you, when you're selling free storage over the Internet, you don't need the ability to write 1GB/sec. Plus they're passing everything through HTTPS.
2. They'll run into a lot of problems with disk drive differences. Since they aren't a big name, they can't negotiate a steady supply of a particular model of drive. Thus, over time, they are subject to the constantly changing consumer drive industry. While you can replace a drive with a different one, the new drive must have at least as many cylinders as the largest remaining drive. Two drives with the same advertised capacity may not have the same number of cylinders. Over time they may find certain arrays increasingly difficult to find drives for. Drives have to be carefully labeled and tracked so replacements can be made correctly if one fails.
3. Because they are using cheap consumer disk drives, the failure rates are about 10x that of industrial grade drives used by high end storage vendors. Furthermore, because there's no proactive monitoring of the drive arrays, the chance of a second drive failure after the first is quite higher than you imagine, because during reconstruction all of the remaining drives have to be read completely, leading to a greater chance of finding a previously undetected error. Reconstruction is a slow process as well, and performance will be degraded during that time.
4. Surprisingly, only something like half of storage server failures are disk drive failures. The other half are things like interconnect failures, power supply failures, etc. These drives aren't dual interconnected and don't have redundant power supplies, so any failure of any other component will take down the entire array. Data won't be lost from a non-drive failure but it will still be unavailable to the customer for a period of time.
I wish these guys luck, but I think one of their primary issues will be customers leaving their service due to reliabilty issues. Those servers will require constant monitoring and maintenance. - wondertwins, on 09/04/2009, -2/+37I was about to say but i excuse him since he comes from Colombia.
- JHW539, on 09/04/2009, -3/+26Storage doesn't have to be scary or expensive. Kudos to these guys for getting that story out there. Personally, I believe a 1 terabyte RAID 1 setup should be the standard for *home* use at this point.
(As a cooling geek, a mechanical engineer who does datacenters, it looks like the fans and airflow could be optimized a bit. That side slotting is strictly for show, right? Side to side airflow makes datacenter designers cry. Even Cisco has finally knocked that crap off.) - 4AntiStupid, on 09/04/2009, -4/+26That's just the hardware cost. It also wouldn't scale down to the needs of most companies because the performance of each unit is pretty crappy with all those disks behind a small pipe. It depends on the highly distributed I/O to be useful.
- Phi01, on 09/04/2009, -3/+25I see what you did there...
- BlitzkrIg, on 09/04/2009, -2/+23Lol, it is quite a bit easier to find people to block when they think they can fool the internets with a name like... 4AntiStupid.
You made my job a bit easier. Thanks, Stupid. - kanojo1969, on 09/04/2009, -1/+22It's probably the least intrusive advertisement I've ever seen. The guy never even mentions what the company does apart from the incredibly vague 'we backup customer's data'. This isn't spam by any definition of the word.
- merreborn, on 09/04/2009, -0/+18Travel back in time to 2006. 2006's digg had stuff like this all the time.
- TnTBass, on 09/04/2009, -2/+20I've been working on a similar system for myself, albeit without the resources these guys have at their disposal to create a custom case, etc.
On this setup it left me wondering exactly how they determine which of the 45 hard drives in the system have actually failed? That could be quite the headache to determine. Plus, when one power supply fails, they lose half the system, or more, depending on which power supply failed.
They can improve it by having a better enclosure for the HD's - Something that is hot swappable with HD lights to indicate a failure. Double up the PSU's and get some redundancy there, and they'll be golden.
That being said, for home use, I'd love to get my hands on everything except the hard drives so I can slowly populate it myself. Not sure I'd use it for any production/business use though, at least not without having two of those for redundancy purposes. They do say though: "Each pod in itself is just a big chunk of raw storage for an inexpensive price; it is not a “solution” in itself." so I suppose they ignore the redundancy issues in a single pod as their redundancy is handled over multiple pods. - JessterKing, on 09/04/2009, -3/+20kudos i really wish i could double bump
- kevinmoore, on 09/04/2009, -0/+16Sorry, I interpreted your message incorrectly. I thought you were saying that digg has been better than /. for technical news.
- Diosjenin, on 09/04/2009, -1/+16They did it to save money on the RAID cards, I think. With a few port multipliers, you can hook up 10-15 disks on a ~$100 RAID card that would ordinarily only be capable of maintaining four disks. Shooting for a card from, say, Areca that could actually handle the 10-15 disks straight would be more on the order of ~$1000.
Your bandwidth suffers, of course, but you also aren't adding an extra 45% onto the cost of the machine. - dazparkour, on 09/04/2009, -0/+15Too true - because of the Slashdot editors were against a digg tech crowed who dugg up tech articles and front paged them QUICKLY, Slashdot was always behind and that's what drew me here from there.
Now I use both because Slashdot is once again ahead.
And has less dupes. - michrech, on 09/04/2009, -0/+15I take it you didn't fully read the article. These devices do *not* have redundant power supplies. One of the power supplies powers a number of the HDD's, the other powers the PC bits and the remainder of the HDD's. One power supply dies, an entire storage array goes with it. The wrong power supply dies, the entire machine dies.
I have a feeling they mitigate these issues by storing the data across multiple pods, though. - TVarmy, on 09/04/2009, -0/+15They also build their own machine shops, using chisels they whittled from rocks outside the warehouse. These guys are pretty hardcore.
- IllBeBack, on 09/05/2009, -0/+15@4AntiStupid:
Why would you want to block something that is full of this much comic genius? You may want to check your ass and remove any sticks you find stuck up there.
How is it that someone that is an idiot could have come up with that PETAFILE line? - bencho, on 09/05/2009, -0/+14^ i see what you did there
- MrDoug, on 09/04/2009, -0/+13I'd buy one...seriously. You can build things cheaper on your own, but sometimes it's just easier to buy what someone else has figured out how to do.
Bravo... - should2, on 09/04/2009, -1/+14It is incorrect to compare per/PB pricing with enterprise level storage appliance especially without any mentions of IOPS results. They really are two different products. Enterprise level are not just JBODs or computers running off the shelf linux with cheap sata controllers. For 10 times the price you get volumes provided from actice/active clusters, many layers of redundancy at transport and disk level, storage virutalization, dedup, block level mirroring, high quality Fibre Channel disks, "real" disk controllers, 4 hour on site support and a hell of a lot more IOPS. Plus having that many computers controlling that amount of storage just sinks your MTBF. While it is a very ingenious and it does provide for a lot of capacity that's all it really does. No matter how much 1 miillion dollars may sound like, it will always be peanuts compared to the value of 1 PB of enterprise data.
- seandfeeney, on 09/04/2009, -7/+20This may be one-tenth the price of EMC or NetApp but you also get one-tenth the reliability, service, management tools etc. This may be feasible for a small company that doesn't have to comply with government mandates on business continuance, back-up / recovery, remote replication. When you buy an array from EMC, you also get customer service that is the best in the business. Also, if you have a symmetrix, and a drive fails, you have full redundancy, most issues are fixed before you ever even know they happen, and unparalleled management tools to simplify back up and recovery. Also, https access only? Are you kidding me? "There is no iSCSI, no NFS, no SQL, no Fibre Channel." You are seriously going to manage petabytes of storage and not utilize fibre??? Data flow is going to take for ever, and god forbid a drive failed or server crashed. Your RPO would be days if not weeks.
- kanojo1969, on 09/04/2009, -0/+13Not enough physical room for a swappable enclosure - I'm guessing that they simply turn off a failed disk and don't bother replacing it. After three years your 45 drives might be down to 40, whereas if you used swappable enclosures you'd probably only fit 30 drives in there in the first place.
Of course, if there's only a few dozen boxes in use, they probably *do* replace failed drives. But as the racks pile up, it get's less and less worthwhile.
I recall reading somewhere that Google does this for entire servers. If one dies, they just strike it from the list and forget about it. It costs more to find it & fix it than it does to buy a new one. - counterplex, on 09/04/2009, -0/+13Commented?? Really? Oh wait...
- JessterKing, on 09/04/2009, -2/+15he meant built omg just let it go
- gunit99, on 09/04/2009, -1/+13I..I just experienced a total nerdgasm.
- nard3456, on 09/05/2009, -0/+12Well played, sir.
- tgc1, on 09/04/2009, -2/+13Grammar Nazi's.... I hate these guys.
- merreborn, on 09/04/2009, -0/+11"No matter how much 1 miillion dollars may sound like, it will always be peanuts compared to the value of 1 PB of enterprise data."
Of course, on the other hand, these guys aren't storing "enterprise data". They offer unlimited online backup for $5/mo. So, their data is worth a lot less than "enterprise data".
So in the end, yes: this would make a TERRIBLE enterprise storage solution. But similarly, enterprise storage solutions are a terrible choice for a company like backblaze. Why should they pay $2 million per PB to store a bunch of backups of cat pictures, porn movies, and malware-laden ***** people downloaded off of limewire? - Jinkley, on 09/04/2009, -3/+13Good. Someone mentioned that IT Crowd episode.
For all who don't know:
http://www.youtube.com/watch?v=fhMLwBHwq1o - Phych, on 09/04/2009, -0/+10There's also no redundancy aside from RAID. If anything other than a hard drive fails, the whole pod is down.
-
Show 51 - 100 of 234 discussions
© Digg Inc. 2009
— Content created and posted by Digg users is