[mapserver-users] Faster MapServer Stand-Alone?

[Paul Ramsey]

Frederik Ramm wrote:

> Please correct me if I'm wrong but I got the impression that,
> basically, the MapServer will always run as a CGI. That means that for
> each request it will not only start a new process but also open and
> parse the map and all shape files. Then, it creates an output file
> which will be read by the web server in order to pass it on to the
> client.
> 
> I haven't run any load tests but it seems obvious that this wastes a
> lot of performance. 

Probably it wastes *some* performance, but before you go and develop a
long-running mapserver at great personal expense, you might want to
consider the relative loads involved in:

1 - starting up the CGI process
2 - parsing the mapfile
3 - finding and reading in the relevant vectors
4 - rasterizing the vectors

Because I have empirically found that mapserver is slowest when I ask it
to rasterize alot of vectors. The more data I actually require *in the
image* in slower it is. And a long-run process is not going to avoid
that requirement.

So if the rasterization (4) is 90% of the CPU requirement, spending your
effort optimizing for 1, 2 and 3 might not be the best investment.

I understand your concerns, it was the first thing which leapt to my
mind when I started installing and understanding mapserver, but my
actual experience *using* it has shown it to be blazingly fast.

-- 
      __
     /
     | Paul Ramsey
     | Refractions Research
     | Email: pramsey at refractions.net
     | Phone: (250) 885-0632
     \_