[Benchmarking] Mapserver Shapefile Results - Poor performance under Load

Smith, Michael ERDC-USACE-NH michael.smith at usace.army.mil
Mon Oct 19 05:27:58 EDT 2009


Frank,

Daniel, Jeff and I are seeing long render times (at least as far as Jmeter
is seeing) when running Mapserver against shapefiles under load. We are not
seeing this under oracle, postgis, or even sde.

You can see it in the numbers listed below.  Once we get above a certain
load level, you see the much larger max times in the results. We can¹t seem
to figure whats causing it. Especially when oracle and sde are reporting
better numbers. 

We are seeing similar values for 5.4 and 5.6b3 and in the large shapefile as
well as the tiled shapefiles.

Any ideas?

Mike

--------------------------------------------------

********
RUN 2
********
Created the tree successfully using mapserver/vector/shp_edges_merge.jmx
Starting the test @ Mon Oct 19 01:36:27 EDT 2009 (1255930587831)
Waiting for possible shutdown message on port 4445
Generate Summary Results +     5 in   1.1s =    4.7/s Avg:   205 Min:   178
Max:   233 Err:     0 (0.00%)
Generate Summary Results +    54 in  10.0s =    5.4/s Avg:   182 Min:   128
Max:   275 Err:     0 (0.00%)
Generate Summary Results =    59 in  11.1s =    5.3/s Avg:   184 Min:   128
Max:   275 Err:     0 (0.00%)
Generate Summary Results +    48 in  10.0s =    4.8/s Avg:   202 Min:   135
Max:   352 Err:     0 (0.00%)
Generate Summary Results =   107 in  21.0s =    5.1/s Avg:   192 Min:   128
Max:   352 Err:     0 (0.00%)
Generate Summary Results +   121 in  10.9s =   11.1/s Avg:   817 Min:   250
Max:  7501 Err:     0 (0.00%)
Generate Summary Results =   228 in  31.1s =    7.3/s Avg:   524 Min:   128
Max:  7501 Err:     0 (0.00%)
Generate Summary Results +   109 in  11.8s =    9.3/s Avg:   845 Min:   138
Max:  4554 Err:     0 (0.00%)
Generate Summary Results =   337 in  41.1s =    8.2/s Avg:   628 Min:   128
Max:  7501 Err:     0 (0.00%)
Generate Summary Results +   125 in  13.5s =    9.2/s Avg:  1458 Min:   451
Max:  4513 Err:     0 (0.00%)
Generate Summary Results =   462 in  51.1s =    9.0/s Avg:   852 Min:   128
Max:  7501 Err:     0 (0.00%)
Generate Summary Results +   118 in  18.5s =    6.4/s Avg:  1522 Min:   275
Max:  9620 Err:     0 (0.00%)
Generate Summary Results =   580 in  61.1s =    9.5/s Avg:   989 Min:   128
Max:  9620 Err:     0 (0.00%)
Generate Summary Results +   121 in  20.7s =    5.8/s Avg:  1371 Min:   173
Max: 16319 Err:     0 (0.00%)
Generate Summary Results =   701 in  71.5s =    9.8/s Avg:  1055 Min:   128
Max: 16319 Err:     0 (0.00%)
Generate Summary Results +   112 in   9.9s =   11.4/s Avg:  2446 Min:   541
Max:  5695 Err:     0 (0.00%)
Generate Summary Results =   813 in  81.1s =   10.0/s Avg:  1246 Min:   128
Max: 16319 Err:     0 (0.00%)
Generate Summary Results +   119 in  18.8s =    6.3/s Avg:  2852 Min:  1219
Max: 16362 Err:     0 (0.00%)
Generate Summary Results =   932 in  91.1s =   10.2/s Avg:  1451 Min:   128
Max: 16362 Err:     0 (0.00%)
Generate Summary Results +   123 in  28.3s =    4.4/s Avg:  2781 Min:   459
Max: 24337 Err:     0 (0.00%)
Generate Summary Results =  1055 in 101.2s =   10.4/s Avg:  1606 Min:   128
Max: 24337 Err:     0 (0.00%)
Generate Summary Results +   121 in  38.9s =    3.1/s Avg:  2694 Min:   270
Max: 29587 Err:     0 (0.00%)
Generate Summary Results =  1176 in 111.1s =   10.6/s Avg:  1718 Min:   128
Max: 29587 Err:     0 (0.00%)
Generate Summary Results +   126 in  33.8s =    3.7/s Avg:  2604 Min:   239
Max: 33130 Err:     0 (0.00%)
Generate Summary Results =  1302 in 121.1s =   10.8/s Avg:  1804 Min:   128
Max: 33130 Err:     0 (0.00%)
Label   Count   Avg     Min     Max     Errors  Throughput
1       100     187     128     279     0       5.3
10      200     725     138     7501    0       11.8
20      400     1438    173     16319   0       11.9
40      618     2638    230     33130   0       12.1

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/benchmarking/attachments/20091019/d89ec2ed/attachment.html


More information about the Benchmarking mailing list