PerformanceTesting
|| '''Description''' || '''Parallel Requests''' || '''Number of request done in 20 seconds''' ||
|| CGI request on state_shp layer with square extents and square maps * || 1 || 297 to 305 ||
|| WMS request on state_shp layer with square extents and square maps * || 1 || 270 to 280 ||
|| CGI request on state_shp layer with square extents and square maps with 100 layer in mapfile ** || 1 || 275 to 280 ||
|| CGI request on state_shp layer with square extents and square maps with EPSG code in map object *** || 1 || 295 to 297 ||
Note that the state_shp layer is a big shapefile (>3,000,000 line features) and is drawn with a 1 pixel black line.
* Used the mapfile below out of the box with the extents below to draw a 500x500 square map with a square extent. I've only tested with one parallel request since I only have one CPU.
** Copied the 3 layers in the mapfile multiple times to get a total of 100 layers. The copied layers were not called like the 3 original ones.
*** Only changed the MAP projection block to be "init=epsg:4326" instead of the projection string.
== Plan of Action ==
- For the EPSG code performance problem, FrankW is planning to implement some caching directly in PROJ.4. More details to come, but on the PROJ.4 side, not a lot directly in MapServer.
- With really large shapefiles, some profiling revealed that a lot of time is passed in msGetBit called by either msShapeFileLayerNextShape and msFilterTreeSearch. We acknowledge that the bit mask is not the optimal solution for very large shapefile (>3M features). the only thing we plan to do right now is to pass from a 8bit mask to a 32 bit mask.
==== Results ====
|| CGI request on state_shp EPSG code in map object and with PROJ.4 fix to cache EPSG file || 1 || 297 to 305 ||
|| CGI request on state_shp with the bitmask fix [http://trac.osgeo.org/mapserver/ticket/2930 ticket 2930] || 1 || 372 to 378 ||
- It seems that the PROJ.4 fix removed most of the overhead of using epsg projection definition.
- ticket 2930 help a lot when reading big shapefiles. In our case, drawing small part of more than 3M features, we got around 20% more speed.
== Test machine ==
Dell Inspiron 6000 laptop
|| '''RAM''' || 2GB ||
|| '''CPU''' || Intel Pentium M 1.73Ghz ||
== Tools ==
MapServer 2008/03/07 trunk versionBR
[http://www.acme.com/software/http_load/ http_load] for load creationBR
gprof 2.18.93.20081009BR
GeoServer performance test data and mapfile (see below for the mapfile)BR
=== Width 500x500 Extent: ===
-97.0+32.65+-96.95+32.7
-97.1+32.5+-97+32.6
-97.06+32.5+-96.96+32.6
-97.07+32.5+-96.97+32.6
-97.09+32.5+-96.99+32.6
-97.1+32.6+-97+32.7
-97.07+32.6+-96.97+32.7
-97.075+32.5+-96.975+32.6
-97.079+32.5+-96.979+32.6
-97.09+32.5+-96.99+32.6
-97.04+32.5+-96.94+32.6
-97.06+32.5+-96.96+32.6
-97.06+32.53+-96.96+32.63
-97.06+32.56+-96.96+32.66
-97.05+32.56+-96.95+32.66
-97.05+32.58+-96.95+32.68
-97.04+32.58+-96.94+32.68
-97.03+32.587+-96.93+32.687
-97.035+32.54+-96.935+32.64
-97.03+32.54+-96.93+32.64
-97.03+32.55+-96.93+32.65
-97.02+32.545+-96.92+32.645
-97.02+32.554+-96.92+32.654
-97.088+32.5+-96.988+32.6
-97.085+32.5+-96.985+32.6
-97.08+32.5+-96.98+32.6
-97.08+32.55+-96.98+32.65
-97.06+32.55+-96.96+32.65
-97.06+32.54+-96.96+32.64
-97.03+32.59+-96.93+32.69
=== bench.map Mapfile ===
# Map server benchmark map file
MAP
IMAGETYPE PNG
EXTENT -180 -90 180 90
SIZE 400 300
SHAPEPATH "./data"
IMAGECOLOR 255 255 255
FONTSET "./fonts/fonts.list"
#PROJECTION
# "init=epsg:4326"
#"proj=longlat"
#"ellps=WGS84"
#"datum=WGS84"
#END
# WMS and WFS metadata
WEB
METADATA
"wms_title" "WMS Demo Server"
"wms_onlineresource" "http://192.168.1.100/~pramsey/cgi-bin/mapserv.fcgi?map=/Users/pramsey/Code/topp-benchmarks/bench.map&"
"wms_srs" "EPSG:4326 EPSG:32632"
"wfs_title" "WFS Demo Server"
"wfs_onlineresource" "http://192.168.1.100/~pramsey/cgi-bin/mapserv.fcgi?map=/Users/pramsey/Code/topp-benchmarks/bench.map&"
"wfs_srs" "EPSG:4326"
END
END
# STATE_POLYGONS
LAYER
NAME states
METADATA
"wms_title" "Distribution of population in USA, polygons"
"wfs_title" "USA population"
"gml_featureid" "ID"
"gml_include_items" "all"
END
DATA states
TYPE POLYGON
LABELITEM "STATE_ABBR"
DUMP TRUE
# Projection
#PROJECTION
# "proj=longlat"
# "ellps=WGS84"
# "datum=WGS84"
#END
CLASS
NAME 'less than 2M'
EXPRESSION ([PERSONS] < 2000000)
STYLE
COLOR 77 255 77
OUTLINECOLOR 0 0 0
END
LABEL
COLOR 0 0 0
ANTIALIAS FALSE
TYPE TRUETYPE
FONT "TimesNewRoman"
SIZE 11
POSITION CC
PARTIALS FALSE
END
END
CLASS
NAME 'between 2M and 4M'
EXPRESSION (2000000 <= [PERSONS] AND [PERSONS] < 4000000 )
STYLE
COLOR 255 77 77
OUTLINECOLOR 0 0 0
END
LABEL
COLOR 0 0 0
ANTIALIAS FALSE
TYPE TRUETYPE
FONT "TimesNewRoman"
SIZE 11
POSITION CC
PARTIALS FALSE
END
END
CLASS
NAME 'more than 4M'
EXPRESSION ([PERSONS] >= 4000000 )
STYLE
COLOR 77 77 255
OUTLINECOLOR 0 0 0
END
LABEL
COLOR 0 0 0
ANTIALIAS FALSE
TYPE TRUETYPE
FONT "TimesNewRoman"
SIZE 11
POSITION CC
PARTIALS FALSE
END
END
END
# Tiger texas, postgis, full roads (1M lines)
LAYER
NAME tiger_shp
METADATA
"wms_title" "Tiger lines, Texas, Shapefile"
"wfs_title" "Tiger lines, Texas, Shapefile"
"gml_featureid" "ID"
"gml_include_items" "all"
"wms_extent" "-106.64395234549875 25.841083405399676 -93.51951316494875 36.50068992305937"
END
DATA tiger_shp
STATUS ON
TYPE LINE
#DUMP TRUE
# Projection (shared by all layers)
#PROJECTION
#"init=epsg:4326"
# #"proj=longlat"
# #"ellps=WGS84"
# #"datum=WGS84"
#END
CLASS
STYLE
COLOR 0 0 0
END
END
END
# Merano TIFF layer (single file, tiled, overviews)
LAYER
NAME merano2
DATA "merano.tiff"
TYPE RASTER
#PROCESSING "RESAMPLE=AVERAGE"
#PROCESSING "RESAMPLE=BILINEAR"
METADATA
"wms_title" "Merano TIFF imagery"
"wms_extent" "663686.6867493344 5171720.999428161 664366.6867493344 5172320.999428161"
END
# Projection
#PROJECTION
# "proj=utm"
# "zone=32"
# "ellps=WGS84"
# "datum=WGS84"
# "units=m"
#END
END
# End of map file
END
