[Mapserver-dev] Fwd: MapServer testing results

Paul Ramsey pramsey at refractions.net
Sat Sep 18 15:57:27 EDT 2004


Howard,
   We are getting some very disturbing results from our ArcIMS vs  
Mapserver testing. The nub of the problem is, ArcIMS is whipping our  
ass.
   The log results below were obtained by putting a timer at the start  
and end of the msLayerOpen function.
   So, we can see that the first layer takes 2750ms to open (wow!), and  
all subsequent layers take very little time (from 0-16ms).
   The place we are getting destroyed is in the actual layer processing,  
retrieving and drawing the data (the difference between the layer  
ending time and starting time of the subsequent layer is the amount of  
time between getting an SDE connection handle and being done with it).
   So it looks like FastCGI alone will not be the cure to our ArcSDE  
theatre of pain.  I will add some more metering this weekend so we can  
see more about where the time is being spent in the long gaps for some  
of these layers.
   In the mean time, any ideas?  On a pure shapefile test, Mapserver has  
a slight performance edge on ArcIMS (10-15%) for our testing, so I do  
not necessarily suspect the rendering (though Brock might want to send  
the test mapfile we are using to this list).  But the ArcSDE  
performance is criminal -- something is amiss.
   Any help, deeply appreciated,
Paul

Begin forwarded message:

> From: Brock Anderson <banders at refractions.net>
> Date: Sat Sep 18, 2004  12:09:56 PM Canada/Pacific
> To: Paul Ramsey <pramsey at refractions.net>
> Subject: MapServer testing results
>
> Hi Paul,
>
> I did some testing on the new copy of mapserver with debugging output.  
>  I have some results that I thought you might like to look at,  
> although I don't know exactly how to interpret them.  I am not sure  
> what happens in the MapServer code between the "layer starting time"  
> and the "layer ending times" labels.  Is that where SDE connections  
> are made?  Also, I assume the "layer time diffs" is measured in  
> miliseconds, but I'm not sure.
>
> I ran this test several times, and these results are typical.  I also  
> ran some tests with other requests of varying complexity.  I can tell  
> you about those on monday if you like.
>
> I made this request to ArcIMS for all the 1:2M layers of the BC  
> Basemap data:
> http://slkapps2.env.gov.bc.ca/wmsconnector/com.esri.wsit.WMSServlet/ 
> TEST_common_service?request=getmap&layers=BC_RIV_LAKE_WET_POLYS_500M,BC 
> _WATER_LINES_500M,BC_TRANSPORT_LINES_500M,BC_PROV_BOUNDARIES_LINES_500M 
> ,BC_MANMADE_LINES_500M,BC_MAJOR_CITIES_POINTS_500M,BC_WATER_POINTS_500M 
> ,BC_TRANSPORT_POINTS_500M,BC_SPOT_ELEVATION_POINTS_500M,BC_MANMADE_POIN 
> TS_500M&version=1.1.1&STYLES=&SRS=EPSG:42102&BBOX=600000,800000,1500000 
> ,1300000&WIDTH=1000&HEIGHT=1000&FORMAT=PNG
> Response time: 15641 ms
>
> For a similar MapServer request:
> http://slkapps2.env.gov.bc.ca/mapserver/ 
> mapserv.exe?map=D:\apps_nt\bcgov\axl\LOPN_test\MapServerCommonService1. 
> map&request=getmap&layers=BC_RIV_LAKE_WET_POLYS_500M,BC_WATER_LINES_500 
> M,BC_TRANSPORT_LINES_500M,BC_PROV_BOUNDARIES_LINES_500M,BC_MANMADE_LINE 
> S_500M,BC_MAJOR_CITIES_POINTS_500M,BC_WATER_POINTS_500M,BC_TRANSPORT_PO 
> INTS_500M,BC_SPOT_ELEVATION_POINTS_500M,BC_MANMADE_POINTS_500M&version= 
> 1.1.1&STYLES=&SRS=EPSG:42102&BBOX=600000,800000,1500000,1300000&WIDTH=1 
> 000&HEIGHT=1000&FORMAT=PNG
> Response time: 96031 ms
>
> Debugging output for the above mapserver request:
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> starting
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_RIV_LAKE_WET_POLYS_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 1140.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 3890.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 2750.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_WATER_LINES_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 7281.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 7328.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 47.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_TRANSPORT_LINES_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 13876.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 13891.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 15.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_PROV_BOUNDARIES_LINES_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 14532.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 14548.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 16.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_MANMADE_LINES_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 14610.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 14610.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 0.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_MAJOR_CITIES_POINTS_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 14688.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 14704.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 16.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_WATER_POINTS_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 14751.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 14767.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 16.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_TRANSPORT_POINTS_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 14798.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 14798.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 0.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_SPOT_ELEVATION_POINTS_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 14845.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 14860.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 15.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting: BC_MANMADE_POINTS_500M
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer starting time: 64771.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer ending times: 64787.000000
> [Sat Sep 18 11:11:20 2004] [error] [client 142.32.21.19] mapserver  
> layer time diffs: 16.000000
>
      Paul Ramsey
      Refractions Research
      Email: pramsey at refractions.net
      Phone: (250) 885-0632




More information about the mapserver-dev mailing list