Web Hosting Talk : Web Hosting Main Forums : Colocation and Data Centers : twin Atom ITX nodes in single mini 1U chassis

[cwl@apaqdigital]

what do you guys think about this Travla C147 Dual Mini ITX 1U chassis:
http://www.e-itx.com/travla-c147.html
the single 15" deep 1U chassis accommodates TWO completely independent hardware nodes in ITX form factor, such as Intel D945GCLF2 Atom 330, with independent PSU.

for high-end configuration, you can also put twin supermicro PDSLM ITX boards in there, and run mobile Core2 duo, 2x 2.5” VelociRaptors per node

[Dougy]

That looks quite interesting

[jayglate]

That is really slick for those who have space as a concern. However as a dedicated server it would be pretty worthless where one clients needs could affect the other ones uptime.

[rghf]

We've got about 80 of those in production. Work well..

[HAClusterTech]

I would definitely be interested in those for some of our projects.

If you plan on offering them, let me know

[cwl@apaqdigital]

Originally Posted by a2b2

We've got about 80 of those in production. Work well..

how are the power supplies holding up? I'm always worry about the long term reliability of non-supermicro made PSU...

[rghf]

Not had a single failure in 8 months

Rus

[cAPTAIN^k]

Wow these looks nice!

[FastServ]

Originally Posted by jayglate

That is really slick for those who have space as a concern. However as a dedicated server it would be pretty worthless where one clients needs could affect the other ones uptime.

Incorrect. It is two completely isolated servers inside. Even have their own power supplies (so the reboot ports won't reboot both nodes, ect)

[PhMatt]

Originally Posted by FastServ

Incorrect. It is two completely isolated servers inside. Even have their own power supplies (so the reboot ports won't reboot both nodes, ect)

I think what Jason was eluding to is if 2 independent customers share the same server, if customer A has a hardware issue (drive failure for example) that requires replacement, customer B on the same server would be inconvenienced as well. But some nice potential for a BOGO offer or redundancy I’d imagine.

[HNLV]

I'm with apaq here, I am weary about anything non-supermicro.

Originally Posted by PhMatt

I think what Jason was eluding to is if 2 independent customers share the same server, if customer A has a hardware issue (drive failure for example) that requires replacement, customer B on the same server would be inconvenienced as well. But some nice potential for a BOGO offer or redundancy I’d imagine.

VERY VERY VERY VERY GOOD POINT!!!!!!

So unless you are running your OWN corporate or personal nodes...these chassis seem to be problematic, not the design aspect but the usage. Definitely not for a dedicated server company cause 1 server being down is bad enough, but TWO???

[cwl@apaqdigital]

Originally Posted by HNLV

I'm with apaq here, I am weary about anything non-supermicro.



VERY VERY VERY VERY GOOD POINT!!!!!!

So unless you are running your OWN corporate or personal nodes...these chassis seem to be problematic, not the design aspect but the usage. Definitely not for a dedicated server company cause 1 server being down is bad enough, but TWO???

if you can install the twin-node server on rack with nice pair of slide rails, then performing hardware service on one node without powering down 2nd node is definitely feasible. of course, you need to plan ahead with long enough power/network cables so that when rails are fully extended, the power/network feed can remain intact for the node with no hardware issues.

[michcolo]

Originally Posted by cwl@apaqdigital

if you can install the twin-node server on rack with nice pair of slide rails, then performing hardware service on one node without powering down 2nd node is definitely feasible. of course, you need to plan ahead with long enough power/network cables so that when rails are fully extended, the power/network feed can remain intact for the node with no hardware issues.

Though no one likes it when the server on the very top of the rack goes down and they have to replace a hard drive while standing on a chair.

[cwl@apaqdigital]

Originally Posted by kdebs

Though no one likes it when the server on the very top of the rack goes down and they have to replace a hard drive while standing on a chair.

no chair, use ladder, please! it may be a bit difficult, but still very feasible.

I think the benefit of 80-nodes capacity per rack outweights the inconvenience of climbing a ladder occasionally.

[michcolo]

Originally Posted by cwl@apaqdigital

no chair, use ladder, please! it may be a bit difficult, but still very feasible.

I think the benefit of 80-nodes capacity per rack outweights the inconvenience of climbing a ladder occasionally.

Haha, yes ladder would be better.

Does anyone make something similar but where the two separate nodes share redundant power supplies? I.E. there's a bus between them that allows one power supply to fail over to the other.