Nathan,

you can do as many stress tests as you want - nodejs beats apache on any
platform
just serving static content and you can't convince me that Apache serves
dynamic
content faster than static cashed content.

And btw, if you want to compare nodejs on a 1 core power 5 you need to
cluster it
into 2 clusters since power5 has two SMT treads per. core and you also have
to make
sure (and that is not so easy) that nodejs actual runs Async and no Sync
that most
example of simple serving on nodejs does.

and just to make a comparison - on a 1 core 2 tread Intel Sindows will
serve aprox.
2.000 req/sec and on Linux aprox. 4.000 req/sec using ab.exe as the tester.

On Wed, Apr 11, 2018 at 9:32 PM, Nathan Andelin <nandelin@xxxxxxxxx> wrote:

Henrik,

Given your last response, I'm not sure that I understand how you are coming
up with 30 ms, or what you might be measuring. It initially appeared to be
attributed to network latency on a wide area network. Now you appear to be
attributing it to establishing and dropping HTTP connections.

I'm reminded of a stress test that I performed on some of our web
applications that generated lists and performed database transactions. I
was running against a single-core Power 5 server, using LoadRunner to flood
the HTTP server with requests for dynamic content, and report the results.

Server-side processing entailed the HTTP server receiving the requests and
forwarding them to RPG programs, which parsed them, and performed DB I/O,
and generated a response, which the HTTP thread returned to the client.

LoadRunner was reporting an average of 600 successful request-response
cycles per second, which included network latency over a LAN. So it's hard
form me to image what you might be doing that causes 30 ms of latency. Are
you talking specifically about XMLSERVICE?





On Wed, Apr 11, 2018 at 12:40 PM, Henrik Rützou <hr@xxxxxxxxxxxx> wrote:

Nathan,

there is a big difference, most browsers sets a kepp-alive value on their
HTTP connection
to 300 seconds - servers that request services on other servers don't do
that as default
so the establish a new HTTP connection for each call to XMLSERVICE and
that
adds
more that 30 ms to the call.

Besides that - there is a big difference on response time running
localhost
and over a
network where you also have to take the network latency into account.

On Wed, Apr 11, 2018 at 8:04 PM, Nathan Andelin <nandelin@xxxxxxxxx>
wrote:

Henrik,


On Wed, Apr 11, 2018 at 11:08 AM, Henrik Rützou <hr@xxxxxxxxxxxx>
wrote:

Nathan,

you are not able to send even a small package over HTTP and network
in
less
than 30 ms...


You're evidently adding network latency in your 30 ms assertion. I was
just
measuring the 1-2 ms time elapsed in the IBM i Apache server in order
to
perform request-response I/O, and saying that small latency could not
be
much different than that of QZDASOINIT interfaces.

Most sites that I visit have about 90 ms of network latency over the
Internet, plus 200-300 ms of latency attributed to request handling. So
the
1-2 ms of time spent by our IBM i Apache servers is a very small
portion
of
overall latency in most web sites.
--
This is the Web Enabling the IBM i (AS/400 and iSeries) (WEB400)
mailing
list
To post a message email: WEB400@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/web400
or email: WEB400-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/web400.




--
Regards,
Henrik Rützou

http://powerEXT.org <http://powerext.org/>
--
This is the Web Enabling the IBM i (AS/400 and iSeries) (WEB400) mailing
list
To post a message email: WEB400@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/web400
or email: WEB400-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/web400.


--
This is the Web Enabling the IBM i (AS/400 and iSeries) (WEB400) mailing
list
To post a message email: WEB400@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/web400
or email: WEB400-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/web400.





As an Amazon Associate we earn from qualifying purchases.

This thread ...

Follow-Ups:
Replies:

Follow On AppleNews
Return to Archive home page | Return to MIDRANGE.COM home page

This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].

Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.