... additionally, it looks like the IP address for the ATK website is
assigned to Deltaland who offers virtual hosting services. So it's
quite possible that ATK is being hosted on a shared server.
... additionally, :-), the website is running PHP 4.3 which is almost 5
years old. There have been MANY changes and improvements since then
that could affect the performance.
In terms of the architecture of the site, it is using the I-Buildings
E-business Platform (
http://atk.achievo.org/product/about). I haven't
used it, but it sounds complicated. :-)
Kevin
-----Original Message-----
From: web400-bounces@xxxxxxxxxxxx [mailto:web400-bounces@xxxxxxxxxxxx]
On Behalf Of Kevin Schroeder
Sent: Friday, September 11, 2009 9:45 AM
To: Web Enabling the AS400 / iSeries
Subject: Re: [WEB400] PHP - Best Appication Structure
A ping isn't a good measurement to use. It represents one packet
whereas a TCP connection requires at least 3 for the handshake. I got
343ms from the first SYN to the last ACK before the response started.
PHP, by default, uses an output buffer so it will not send any packets
to the client until the request has finished executing, or the output
buffer has flushed. If I measure the time from last ACK to the first
response packet I get 258ms. That would be a more accurate measurement
of what is going on their machine. I might need to add 84ms (168ms/2
assuming equal bi-directional latency) to account for latency bringing
the possible response time up to 342ms. But even that has way too much
variability in it.
In order for this to be a proper apples-apples comparison you would need
to run what they're running on your system with their data and then
compare the two. Perhaps you did and I missed the message but there
seems to be a lot of unknowns. Clearly simple PHP will run faster than
complex PHP but as I had noted in my previous email, the likelihood of a
250ms response affecting you more than a 10ms response is very low.
From the end-user point of view, nothing below 500ms will make much of
a
difference, especially if there are images and JavaScript.
Kevin
-----Original Message-----
From: web400-bounces@xxxxxxxxxxxx [mailto:web400-bounces@xxxxxxxxxxxx]
On Behalf Of Nathan Andelin
Sent: Friday, September 11, 2009 9:12 AM
To: Web Enabling the AS400 / iSeries
Subject: Re: [WEB400] PHP - Best Appication Structure
From: Thorbjoern Ravn Andersen
Perhaps this has not been done fr the ATK installation?
Netcraft indicates that
http://atk.achievo.org is running on Linux in
Denmark. When I ping the server, the response takes 194 milliseconds,
as opposed to just 1 millisecond to ping my server. That's more network
latency than I originally estimated over the internet.
By removing network latency, it appears that time (with rounding) spent
processing a record change on the server would be more like:
900 + 500 - 200 - 200 = 1000 milliseconds
vs.
3 + 2 - 1 - 1 = 3 milliseconds
The following screen-shot compares the two applications running within
my portal.
http://www.radile.com/rdweb/temp/atkdemo.html
Other than Netcraft indicating Linux, I don't know what sort of hardware
might be supporting the ATK demo. But OnePoint portal is running on a
1-cpu, 1-core, 1-thread, 1MB-RAM, 0-L3 cache model 520 server.
-Nathan.
midrange.com
