Linux on 390 Port <[email protected]> írta 2005.07.11 09:27:11 
időpontban:

> On the other hand, one accidental slip of the finger in a combined 
services
> configuration and you take down multiple services. It's the principle of 
not
> putting all your eggs in one basket -- if there is no artificial 
scarcity of
> "hardware", there is no reason to combine services because you can 
actually
> deliver one server, one application at a reasonable cost point with
> acceptable performance, and you don't risk breaking other services
> accidentally.

We - mainframe users - are using mainframes for eg. because of 99.9% 
uptime. Do everyone have a 2nd one as a disaster recovery system - or we 
just put everything in one basket? As Murphy said... 

I understand what you mean, I can imagine a complex system where this can 
be a big problem. I was thinking in a lots of users and not so high number 
of services.


> Yes, not screamingly fast, but reasonable, and how many
> applications *really* need 3 Ghz CPUs for anything but bragging rights? 
If
> your application *really* needs that kind of horsepower, you're on the 
wrong
> system anyway -- you need big Intel or PPC engines, not zSeries.

for eg. SAP has no limits when we are talking about HW needs :), but we 
need the no downtime feature of the z/Series. The company has here 
mainframes since the punch card times, MVS production apps designed to run 
in this environment. Since that times Intel grew up, (and to tell the 
truth, many mission critical apps are running happily on such machines 
too, maybe more than on mainframes), new application developers do not 
care about cpu cycles, but at the end, the user wants short response time. 
Thanks to the java simple stupid apps will need more and more cpu.

> Some problems really *do* need big monolithic LPARs (although that's 
where I
> start to think that it's probably the wrong platform for the app unless 
it
> really is written to exploit zSeries hw features). Most don't, and 
managing
> LPARs is a LOT more expensive in people time than managing virtual 
machines
> under VM. Computer capacity is a lot less expensive than humans these 
days.
> I'll gladly buy more CPU resource if I don't have to buy more people.

This is true only in the happy west side of the world, here just the 
maintenance costs of a z/VM is a good salary..

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to