From: "Rich" 

This is a multi-part message in MIME format.

------=_NextPart_000_034E_01C31D3F.EAF696C0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

   The startup improvements are distinct from the steady state =
improvements.

   Independent from IIS, I/O has improved.  I don't remember any =
specifics.

Rich

  "Geo."  wrote in message =
news:3ec7cd0a$1{at}w3.nls.net...
  "Rich"  wrote in message news:3ec7c573$1{at}w3.nls.net...
  >>   It is much more than that.  I haven't see direct comparisons but =
I
  pretty sure it is faster than IIS 5.0 with the kernel mode cache that =
was
  available.  It's more functional as the old cache was static content =
only.
  If you look at the benchmarks you will see dynamic content performance =
has
  increased.<<

  I saw that in the graphs, I also know they separated the front =
end/back end
  processes somehow which is supposed to improve performance in some
  situations. They also changed the startup resource allocation which =
for a
  machine hosting hundreds of sites can be significant.

  Any idea if disk performance has been enhanced? I noted they used =
striping
  for the benchmark so it's obvious they realized this was a bottleneck =
and
  may have tried to tweak it. I found disk was my main bottleneck and on
  machines with lots of spare cpu using disk compression could improve =
the
  performance of the sites. Seems to me this should be even more =
pronounced
  now due to the changed http.sys setup.

  Geo.


------=_NextPart_000_034E_01C31D3F.EAF696C0
Content-Type: text/html;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable








   The
startup improvements =
are distinct=20
from the steady state improvements.
 
  
Independent from IIS, I/O =
has=20
improved.  I don't remember any specifics.
 
Rich
 

  "Geo." <georger{at}nls.net>">mailto:georger{at}nls.net">georger{at}nls.net>
wrote=20
  in message news:3ec7cd0a$1{at}w3.nls.net..."Rich"=20
  <{at}> wrote in message news:3ec7c573$1{at}w3.nls.net...=
>>  =20
  It is much more than that.  I haven't see direct comparisons but=20
  Ipretty sure it is faster than IIS 5.0 with the kernel mode cache =
that=20
  wasavailable.  It's more functional as the old cache was =
static=20
  content only.If you look at the benchmarks you will see dynamic =
content=20
  performance hasincreased.<<I
saw that in the graphs, =
I also=20
  know they separated the front end/back endprocesses somehow which =
is=20
  supposed to improve performance in somesituations. They also =
changed the=20
  startup resource allocation which for amachine hosting hundreds of =
sites=20
  can be significant.Any idea if disk performance has been =
enhanced? I=20
  noted they used stripingfor the benchmark so it's obvious they =
realized=20
  this was a bottleneck andmay have tried to tweak it. I found disk =
was my=20
  main bottleneck and onmachines with lots of spare cpu using disk=20
  compression could improve theperformance of the sites. Seems to me =
this=20
  should be even more pronouncednow due to the changed http.sys=20
 
setup.Geo.

------=_NextPart_000_034E_01C31D3F.EAF696C0--

--- BBBS/NT v4.01 Flag-4