Jump to content
Sign in to follow this  
mark84

67 terabytes for only $7869

Recommended Posts

With all that storage, I hope they're not using onboard ethernet NICs to hook it all up...

 

Brains: The "shitbox" is less than 1/6 the price if the thingie at your link... If you were in a pinch, it'd do the trick...

Edited by SquallStrife

Share this post


Link to post
Share on other sites

I think that's the first time if seen drives mounted vertically in a rack server (mind you I don't look at rack servers much).

Yeah, not quite. Sun have been doing it for about 4 years.

 

z

Share this post


Link to post
Share on other sites

zebra o god of things that store 1's and 0's

 

does that get your thumbs up or down? lol

Down and up, actually.

 

Fail: Poor quality controllers, prone to errors and parity failure.

Fail: Poor choice of filesystem, based upon WELL KNOWN extents growth bugs in JFS.

Fail: Doesn't scrub/dynamically checksum data on CoW. If you write bad data, because of crappy controllers, you'd never know.

Fail: If you need to rebuild, you're in for a MASSIVE system performance hit, given the way the controllers are laid out, and their entire lack of useful cache.

 

Win: It's bloody cheap.

Win: It's hella modular.

 

I'd never consider using things like this, for critical storage environments. Arguably, a cloud isn't "critical", and by putting stuff in public clouds, people are aware of the dangers.

Share this post


Link to post
Share on other sites

So if they dumped an extra $4-10k or something into it for some decent controllers with cache and changed the file system (ZFS?) it'd be a both thumbs up?

 

Things so cheap you could spend quite a bit more on better hardware and still come out way cheaper than the rest (presumably).

Share this post


Link to post
Share on other sites

So if they dumped an extra $4-10k or something into it for some decent controllers with cache and changed the file system (ZFS?) it'd be a both thumbs up?

For the most part, yeah!

Share this post


Link to post
Share on other sites

Fail: Poor choice of filesystem, based upon WELL KNOWN extents growth bugs in JFS.

Fail: Doesn't scrub/dynamically checksum data on CoW. If you write bad data, because of crappy controllers, you'd never know.

which non zfs filesystems offer these features? or do none? :(

Share this post


Link to post
Share on other sites

Fail: Poor choice of filesystem, based upon WELL KNOWN extents growth bugs in JFS.

Fail: Doesn't scrub/dynamically checksum data on CoW. If you write bad data, because of crappy controllers, you'd never know.

which non zfs filesystems offer these features? or do none? :(

 

It doesn't matter, you can put other filesystems (NTFS, ext3, etc) in ZFS containers. :)

Share this post


Link to post
Share on other sites

It doesn't matter, you can put other filesystems (NTFS, ext3, etc) in ZFS containers. :)

Exactly :). You win a prize for forward-thinking.

 

z

 

NB: Prize may not actually be a prize.

Share this post


Link to post
Share on other sites

i know im hijacking but if the base is mdadm raid5 then any filesystem ontop of it wont protect it against the checksum dynamic scrub whoosit whoopla?

Share this post


Link to post
Share on other sites

i know im hijacking but if the base is mdadm raid5 then any filesystem ontop of it wont protect it against the checksum dynamic scrub whoosit whoopla?

Correct, and hence, we don't use mdadm at all, if we want serious safety in such a situation.

 

z

Share this post


Link to post
Share on other sites

Yes it's cheap and nasty, but you know what? It's showing a shift in enterprise storage away from feature rich, proprietary, expensive units, to feature rich, open source, inexpensive units.

 

 

You know that once ZFS (or something like it) is banged into some SAN firmware with L2ARC type capabilities and the appropriate off the shelf hardware, you've got yourself a high performance, low cost SAN without worrying about having to run OpenSolaris/*BSD.

 

I think we are seeing the start of something here.

Share this post


Link to post
Share on other sites

I would not imagine the noise of this frankenserver would be pleasing in a home environment.

 

Just run Solaris - you know you want to. :)

 

I think what you mean is RAIDZ and Z2 as part of the raid cards firmware. I don't think that can happen since RAIDZ is an integral part of the ZFS filesystem, that is to say i dont think its possible for raidz to exist on its own away from the ZFS filesystem, which is a bit of a wall to that plan.

 

But then I don't consider it a wall either. since we know that the Sun 7000 series unified storage boxes are bloody good -mmmm amber road. I digress.

The combination of Sata drives with SSD's as cache devices with the non fussy loves everything ZFS filesystem is pure win. All of these things go together to make pure awesome. The same would not be possible with other filesystems because they want for expensive disks. Well that's my take on it anyway.

Share this post


Link to post
Share on other sites

They have a point, if someone posted a picture of their car, lets say it was some new falcon model, then you would reply "what a shitbox get a bugatti veyron"

 

Incomparable.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×