[Mapserver-users] performance information

Jerl Simpson jerl.simpson at rkiq.com
Tue Jul 20 14:56:37 EDT 2004


Ed,

Thanks for the reply.
I was afraid it would be this type of discussion.  Too many variables.

Here's the deal.

Mapserver on a freeBSD install...two in fact.
Currently the machines I have to work with are dual Pentium 1.2GHz 
machines with 768Mb of ram.  The machines are COMPAQ DL380 G2 servers. 
Raid 0 with 6 18.2Gb SCSI drives.
Now, we will eventually buy some REAL servers for this function, but for 
now this is what I have to work with.

Initially we won't be doing anything too terribly fancy...and most 
likely not long term fancy either.

I am displaying data for the US currently.  The data is simply political 
boundaries, cities, hydrology, counties, roads, railroads.  All this 
comes out of the ESRI shape files...which I understand are free to 
use???  I need to make sure of this before I make this a production server.


I am sure there is more information needed...so let me know and I'll 
fill in the holes.

Thank you,

Jerl


Ed McNierney wrote:

> Jerl -
> 
> Unfortunately, the only possible answers are (a) a complex discussion or
> (b) blah, blah, blah.
> 
> The performance of any Web map server will depend on a long list of
> factors - the number of users, the frequency of map requests, the size
> of each map request, the average Internet bandwidth to each user, the
> pixel depth and image format served, the amount of RAM on the server,
> speed and number of CPUs, the interface used for hard disks, the number
> of hard disks, their seek times, the number of independent data
> channels, the number of data layers used, the format of each layer
> (raster, vector, database), the file formats used for each data storage
> format, blah, blah, blah.
> 
> The only simple answers are incorrect answers.  If I only need to serve
> one map request to each user, and I can take as long as I like to do it,
> I can serve an almost unlimited number of simultaneous connections
> "without major impact to the system".
> 
> This is a very complicated discussion.  I have been very pleased with
> MapServer's performance - I switched from a single-purpose, dedicated
> raster map server that I wrote myself, to MapServer.  But I worked with
> MapServer and my data for a long time so I understood how to design my
> system (that's something I'd do with any server, not just MapServer).
> 
> Most folks feel I run a "pretty big" MapServer installation.  I serve
> over half a million maps a day to up to around 1,000 simultaneous users
> (more when there's a California earthquake <g>) from about 21 terabytes
> of vector and raster imagery in a variety of formats.  I don't claim to
> be the biggest MapServer installation around, but I certainly serve more
> maps from more data to more users than the average site, and MapServer's
> been great for doing that.  But one could easily choke a system that's
> much "smaller" in several dimensions with bad design.
> 
> If you have a particular MapServer application in mind, there are
> several folks here who could give you advice if you could describe it in
> as much detail as possible.
> 
> 	- Ed
> 
> Ed McNierney
> President and Chief Mapmaker
> TopoZone.com
> -----Original Message-----
> From: mapserver-users-admin at lists.gis.umn.edu
> [mailto:mapserver-users-admin at lists.gis.umn.edu] On Behalf Of Jerl
> Simpson
> Sent: Monday, July 19, 2004 5:15 PM
> To: mapserver-users at lists.gis.umn.edu
> Subject: [Mapserver-users] performance information
> 
> Hello all,
> 
> I am curious if anyone has done stress and performance testing on
> mapserver?
> 
> I know I have seen some numbers, but I can't find those now.
> I am basically wanting to know number of simultaneous connections that
> can be served with out major impact to the system.
> I know this number varies based on the strength of the architecture and
> such.  But if I could find info that says "On my Dual Xeon blah blah
> with 2Gb RAM blah blah"
> 
> Or something like, "I noticed a performance hit when we had X
> simultaneous connections."
> 
> Does anyone have this data?  Or know where I might find such data?
> 
> Thanks very much.
> 
> jerl
> _______________________________________________
> Mapserver-users mailing list
> Mapserver-users at lists.gis.umn.edu
> http://lists.gis.umn.edu/mailman/listinfo/mapserver-users
> .
> 



More information about the mapserver-users mailing list