Re: [tlug] Re: Talk about fast HTTP

[Curt Sampson <cjs@example.com>]

On 2008-10-18 12:35 +0200 (Sat), Christian Horn wrote:

> That usecase could be handled nowadays with an x86-server...

Well, of course. It already is handled with x86 servers!

> ...just seen that HP offers those nowadays, carrying up to 512gb of
> ram which could be used for caching here.

I suppose that's an alternative solution, at probably ten times the
price or more. I may need two or three times the number of servers, but
on the other hand, each of my servers has only 4-8 GB of memory.

The point of my post was not that the problem can't be handled; the
point of my post was: adding network capacity without adding disk I/O
won't fix the problem.

It's more common, when setting up 1U servers, to configure them either
with a single disk or two-disks in a mirrored configuration. Both are
wrong in this case. You can only know that if you know that disk IO is
your bottleneck.

On 2008-10-18 21:28 +0900 (Sat), Bruno Raoult wrote:

> The day you can help banks to do better, you are rich, lucky you!

I think if you want to become rich helping someone, you'd be far better
to work with traders than back-room folks.

cjs
-- 
Curt Sampson       <cjs@example.com>        +81 90 7737 2974   
Mobile sites and software consulting: http://www.starling-software.com