RE: Benchmarks should be better

2001-02-05 Thread Michael J. Cannon

Re: Security (run as root).

In Unix/Linux, you don't have to.

A guide to how to change this is posted at:

http://www.orionsupport.com/articles/unixprocess.html



-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of Michael Quinn
Sent: Sunday, February 04, 2001 2:52 AM
To: Orion-Interest
Subject: RE: Benchmarks should be better


Just a little update..
I have completed benchmarks for the initial apache default index.html.en
with 4 different http servers on Redhat 6.2 with AMD Thunderbird 650 and
128MB Ram.
Apache 1.3.14
Tomcat 3.2
Weblogic 5.1
Orion 1.4.5

I couldn't believe my eyes when Orion served the static html faster than all
of them (Beating apache by about a 30% margin using 20 threads with 5
sockets per thread.  Another amazing thing was resource usage..  The highest
CPU Usage I saw from Orion during the test was a mere 30 percent.  If that
is not a convincing factor, then I don't know what is.

I still have some concerns about the security of orion (running on port 80
as root), and Apache is by nature a more memory intensive application.  It
uses a multi-process -vs- orion's multithreaded technique.  Each process
uses a few megs of memory, so for a large request base, apache will need
more memory so it isn't getting page faults all over the place. ( I am also
running oracle on this box which uses tonnes of memory as many know )
I'll slap in another 128M and see what happens.

Any comments would be appreciated.

Michael Quinn
Software Engineer

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of Michael Quinn
Sent: Friday, February 02, 2001 11:19 AM
To: Orion-Interest
Subject: Benchmarks should be better


Hey all,

I was checking out the benchmarks on www.orionserver.com and they are quite
interesting.

There are a couple of things I am wondering, however.  How does Orion
serving dynamic content pit itself -vs- apache serving static content or
mod_perl stuff.
It would be nice to take the other "sucky" java servers out of the picture
and see a baseline comparison of Orion -vs- Apache on a lot of different
scenarios.

On a side note, can somebody forward me some performance comparisons from
weblogic?  I know they can't be posted on the website, but I would like to
see them.

The reason I ask is of the following importance:

I see a lot of job postings for knowledge of Weblogic.  And about 50% of
telephone interviews ask about it, or bring it up.
I want to know how it performs.

I'm thinking about setting up Weblogic with Apache and Orion, and doing some
performance comparisons which I will be glad to share with everyone.

Thanks for your interest,

Michael


_
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com


_
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com






RE: Benchmarks should be better

2001-02-04 Thread Mark Bernardinis

On Sun, 4 Feb 2001, Michael Quinn wrote:

 Just a little update..
 I have completed benchmarks for the initial apache default index.html.en
 with 4 different http servers on Redhat 6.2 with AMD Thunderbird 650 and
 128MB Ram.
 Apache 1.3.14
 Tomcat 3.2
 Weblogic 5.1
 Orion 1.4.5

 I couldn't believe my eyes when Orion served the static html faster than all
 of them (Beating apache by about a 30% margin using 20 threads with 5
 sockets per thread.  Another amazing thing was resource usage..  The highest
 CPU Usage I saw from Orion during the test was a mere 30 percent.  If that
 is not a convincing factor, then I don't know what is.

 I still have some concerns about the security of orion (running on port 80
 as root), and Apache is by nature a more memory intensive application.  It

Running Orion as root is not recommended. There have been tutorials
available from www.orionsupport.com that recommend better alternatives
such as running Orion from port 8080 or whatever and using port
redirection. This is what I currently use and I can run Orion as the orion
user which I have created.

I am pleased with Orion's results in CPU usage and memory. Keep the bench
marks coming.

 uses a multi-process -vs- orion's multithreaded technique.  Each process
 uses a few megs of memory, so for a large request base, apache will need
 more memory so it isn't getting page faults all over the place. ( I am also
 running oracle on this box which uses tonnes of memory as many know )
 I'll slap in another 128M and see what happens.

 Any comments would be appreciated.

 Michael Quinn
 Software Engineer

 -Original Message-
 From: [EMAIL PROTECTED]
 [mailto:[EMAIL PROTECTED]]On Behalf Of Michael Quinn
 Sent: Friday, February 02, 2001 11:19 AM
 To: Orion-Interest
 Subject: Benchmarks should be better


 Hey all,

 I was checking out the benchmarks on www.orionserver.com and they are quite
 interesting.

 There are a couple of things I am wondering, however.  How does Orion
 serving dynamic content pit itself -vs- apache serving static content or
 mod_perl stuff.
 It would be nice to take the other "sucky" java servers out of the picture
 and see a baseline comparison of Orion -vs- Apache on a lot of different
 scenarios.

 On a side note, can somebody forward me some performance comparisons from
 weblogic?  I know they can't be posted on the website, but I would like to
 see them.

 The reason I ask is of the following importance:

 I see a lot of job postings for knowledge of Weblogic.  And about 50% of
 telephone interviews ask about it, or bring it up.
 I want to know how it performs.

 I'm thinking about setting up Weblogic with Apache and Orion, and doing some
 performance comparisons which I will be glad to share with everyone.

 Thanks for your interest,

 Michael


 _
 Do You Yahoo!?
 Get your free @yahoo.com address at http://mail.yahoo.com


 _
 Do You Yahoo!?
 Get your free @yahoo.com address at http://mail.yahoo.com








Re: Benchmarks should be better

2001-02-04 Thread Burr Sutter

how many requests/responses were simulated?

- Original Message -
From: Michael Quinn [EMAIL PROTECTED]
To: Orion-Interest [EMAIL PROTECTED]
Sent: Sunday, February 04, 2001 3:52 AM
Subject: RE: Benchmarks should be better


 Just a little update..
 I have completed benchmarks for the initial apache default index.html.en
 with 4 different http servers on Redhat 6.2 with AMD Thunderbird 650 and
 128MB Ram.
 Apache 1.3.14
 Tomcat 3.2
 Weblogic 5.1
 Orion 1.4.5

 I couldn't believe my eyes when Orion served the static html faster than
all
 of them (Beating apache by about a 30% margin using 20 threads with 5
 sockets per thread.  Another amazing thing was resource usage..  The
highest
 CPU Usage I saw from Orion during the test was a mere 30 percent.  If that
 is not a convincing factor, then I don't know what is.

 I still have some concerns about the security of orion (running on port 80
 as root), and Apache is by nature a more memory intensive application.  It
 uses a multi-process -vs- orion's multithreaded technique.  Each process
 uses a few megs of memory, so for a large request base, apache will need
 more memory so it isn't getting page faults all over the place. ( I am
also
 running oracle on this box which uses tonnes of memory as many know )
 I'll slap in another 128M and see what happens.

 Any comments would be appreciated.

 Michael Quinn
 Software Engineer

 -Original Message-
 From: [EMAIL PROTECTED]
 [mailto:[EMAIL PROTECTED]]On Behalf Of Michael Quinn
 Sent: Friday, February 02, 2001 11:19 AM
 To: Orion-Interest
 Subject: Benchmarks should be better


 Hey all,

 I was checking out the benchmarks on www.orionserver.com and they are
quite
 interesting.

 There are a couple of things I am wondering, however.  How does Orion
 serving dynamic content pit itself -vs- apache serving static content or
 mod_perl stuff.
 It would be nice to take the other "sucky" java servers out of the picture
 and see a baseline comparison of Orion -vs- Apache on a lot of different
 scenarios.

 On a side note, can somebody forward me some performance comparisons from
 weblogic?  I know they can't be posted on the website, but I would like to
 see them.

 The reason I ask is of the following importance:

 I see a lot of job postings for knowledge of Weblogic.  And about 50% of
 telephone interviews ask about it, or bring it up.
 I want to know how it performs.

 I'm thinking about setting up Weblogic with Apache and Orion, and doing
some
 performance comparisons which I will be glad to share with everyone.

 Thanks for your interest,

 Michael


 _
 Do You Yahoo!?
 Get your free @yahoo.com address at http://mail.yahoo.com


 _
 Do You Yahoo!?
 Get your free @yahoo.com address at http://mail.yahoo.com