Performance Impact of ASP.NET 1.1 and 2.0 Side by Side

[Rich]

Hi,

I know HOW to run these side by side. But what is the performance impact???
Microsoft make a big deal about how it is supported, but provide no
guidelines as to potential performace or capacity considerations.

Is it much greater than the cost of an additional application pool of the
same version?

Anybody have any real world experience, or know of any metrics?

I've found it really hard to find info on this.

Thanks
Rich

 

[Tim Mackey]

hi rich,
my server runs side-by-side with 1.1 and 2.0, and i run a bunch of asp.net
web sites on it. one performance impact i can think of is that you have a
pair of w3wp.exe processes instead of just one. on my server at the moment,
one is 150Mb and the other is taking up 250Mb, but that obviously depends on
traffic and the implementation of your site, session / cache etc. i can't
imagine having 2 processes makes a whole lot of difference because the
footprint of the process itself is comparatively small. whether you rack up
memory usage in one 'big' process or two smaller ones it probably doesn't
make much difference.
you never know, it might even be faster to have two separate processes, with
greater room for parallelisation in a multi-core CPU. that's a very
unqualified opinion though!

tim

 

[Thomas Hansen]

Hi,

I know HOW to run these side by side. But what is the performance impact???
Microsoft make a big deal about how it is supported, but provide no
guidelines as to potential performace or capacity considerations.

Is it much greater than the cost of an additional application pool of the
same version?

Anybody have any real world experience, or know of any metrics?

First of all if you have installed Service Pack 2 from Microsoft it is
ILLEAGAL to publish benchmark results on e.g. .Net Fr.

Though I've heard some rumours about Reflection being actually slower
in 2.0 compared to 1.x
I think most of the other stuff though would be faster...!

Off course if you use delegates (strongly types reflection it'll be a
ZILLION times faster...!!)
The comparison had the exact same code for the Reflection parts...

..t

 

[Tim Mackey]

First of all if you have installed Service Pack 2 from Microsoft it is
ILLEAGAL to publish benchmark results on e.g. .Net Fr.

sounds bizarre, where did you hear that? also, service pack 2 of what,
exactly? Service Pack 2 from Microsoft could almost refer to anything.

tim

 

[Thomas Hansen]

exactly? Service Pack 2 from Microsoft could almost refer to anything.

Sorry..
Windows XP SP2...
Read the EULA of Windows XP SP2...

..t

 

[Rich]

Thanks for that.

Probably looking at putting in its own app pool anyway, (sharing a box with
some stuff that just runs and we dont want to touch) so would get the
overhead of the additional process regardless of the version.

 

[Tim Mackey]

hi thomas,
ah i see. just out of curiosity, i did read the xp pro sp2 EULA from
http://www.microsoft.com/windowsxp/sp2/proeula.mspx
and it makes no restrictions on publishing anything, or benchmarking. my
impression is that Microsoft would encourage the wider developer community
to rigorously test and discuss the performance of their products and suggest
enhancements for future versions. this is the attitude i perceive in recent
years from microsoft employees in general. just look at the Vista beta
program, Microsoft conducted the biggest ever public beta test in history
(AFAIK) which has undoubtedly helped build the best OS in history, in my
humblest opinion i actually like using my computer now since vista,
this is a first, and yes i have used many other OSes so i do have something
to compare against.

it sounds like you may have inadvertendly picked up on some inaccurate anti
MS-propaganda
tim