Re: [cobalt-users] Re: cobalt-users digest, Vol 1 #279 - 20 msgs

[Will DeHaan <will@xxxxxxxxxx>]

Kris Dahl wrote:
> 
> The reason why you go with a HP Netserver (lets take the LH3 for example) is
> obvious.  On a LH3, I can get the following: Rundant power supplies,
> redundant fans, redundant raid controllers, multiple global hot-spares,
> striped RAID, ECC scrubbing memory, etc..  That means it can survive a
> failure of: multiple multiple power supply failures, fans, RAID controllers,
> numerous hard drives, single bit (and double bit) memory errors.  Lets say
> that for some reason it has a triple bit memory error, or the NOS crashes.
> It'll reboot itself after automitically when it detects a hang. 

The RaQ 3 is equipped with a device (watchdog timer) that will reboot
the server if the kernel locks up.  Server availability and reliability
is a top concern for those of us building software. 


	-- Will

> And lets
> say that some data got corrupted--I can dial into the thing remotely and
> reinstall the OS, if need be (would be tricky, but theoretically possible).
> The thing's a rock.
> 
> The Qube offers no availability options.  If you're application would cost
> you thousands of dollars a minute, and you haven't addressed this issue, I
> wouldn't be worried to much about your clients and partners.  I'd be worried
> about your IS staff and the priorities they place on the bottom line.
> 
> The above mentioned Qube is about $1500.  The above mentioned LH3 is
> probably around $20,000.  That is the difference between a mission critical
> enterprise server and a Cobalt server appliance.  It isn't that Cobalt
> hasn't thought about these issues--its that the decided not to build the
> product.  They chose to build a very affordable, very capable server
> appliance.
> 
> -k
> 
> _______________________________________________
> cobalt-users mailing list
> cobalt-users@xxxxxxxxxxxxxxx
> http://list.cobalt.com/mailman/listinfo/cobalt-users