Blade

[Richard Ford]

Using 1955's they are great.

Insanely fast - integrated chassis management where you can see power  
draw and usage and set thermal/power limits.  WHole chassis and  
blades auto step down to conserve power when idle or main power  
supplies suffer failure (non redundant, 3+1 or 2+2 redundant).

Digital KVM seamlessly ingrates with other avocet units for tiered  
KVM usage.

Integrated switches make cable and system management a breeze.

All in all fast and really easy to manage and efficient on power.

We went with blades because here in China, we were spending 30K RMB  
(7.7RM = 1 USD) for a redundant, drac enabled, redundant power  
1850/1950.  A fast SC1425 or SC1435 came in at around 19K with no  
redundancy or management.

A full spec blade comes in at 22K and is therefore more manageable  
than a SC1425/35 and cheaper than a 1950.

By the time the chassis is filled the savings in money - not to  
mention actual management ease and rack space, cable clutter is great.

Also once a chassis is bedded down and setup (do so with full racks  
at a time - once off with foreign engineers) - adding new servers  
literally is a 2 min job.  Any old local employee can then do it (so  
we save good labour charges) and then the better engineers can KVM in  
and setup.  Literally, make an order with Dell, it arrives at the IDC  
and then some monkey slots it in!  Talk about efficiency!

Makes it even more cost effective for our clients based in USA/ 
Australia that off shore here - they don't have to pay good money for  
space, power or server techs that ultimately are a waste of money to  
pay for just putting racks in servers and plugging cables about - can  
spend that money on business analysts and real IT process/UML  
designers.  Blades will enable utility computing and the  
centralization of infrastructure where it deserves the expense that  
it truly does (China/India) considering the actual expertise in  
server setup needed relative to western computer services wage  
demands.... :-s    Blades =really= do slash management costs and  
setup complexity.

Major issue is that the new 1955 use 2.5 inch sas.  So it is small,  
limited to RAID1 or 0 and the size tops out currently at 73GB.  So  
fast local storage at a good size is not possible - have to use  
slower FC storage.  Also due to spindle limitations does not make a  
good DB server.

We use them with vmware to run virtual thin client application  
servers.  One blade with 2 x 2.0GHZ/1333 Xeons with 4GB ram runs 6  
W2K3 terminal services with 2GB allocated ram each and 25 desktop  
users over WAN each and the server barely registers more than 4000MHZ  
in usage - so half the available speed.

If dense ram chips were not so expensive we would run more vm's per  
blade - but given 2GB and 4GB ram prices it is cheaper to buy another  
blade with 1GB DImms.

So in short, use blades, stop wasting time and money on the "lego" of  
computer services and get back to making great applications, process  
designs and happy users with highly available systems with six nines  
uptime.

Cheers,
RF.




On 11 Mar 2007, at 3:05 AM, Mad Unix wrote:

> Whhat kind of blade experiences do you have people?(discontinued,  
> power heating, performance, stability, connection FC SAN, drivers,  
> Linux RH support, $$$$)
> I personally think they have some how limitation, and prefer to  
> stay with 1U or 2U servers
>
> youe VOTE in this issue whould be really appreciated
> -- 
> madunix
> _______________________________________________
> Linux-PowerEdge mailing list
> Linux-PowerEdge at dell.com
> http://lists.us.dell.com/mailman/listinfo/linux-poweredge
> Please read the FAQ at http://lists.us.dell.com/faq