[mapguide-users] MgCooker: bad performance, I can't use it. Help!!!

Franz1965 fsozzi at INTERCAD.CH
Fri May 11 13:13:15 EDT 2012

Hi list,

I have a very big problems on tiles generatig by MgCooker. I've made many
tests in the last months with several projects and different machines, and
the conclusion is that is nearly impossible to use, or you can use if you
can wait days or weeks for finishing processes. In my case , I know nobody
that can wait for so much on updating his data.

I would like to share my experience with somebody that really uses MgCooker
for tiles generation, just to understand if it is really possible to use it
and at which conditions. 

Today I've opened a case ( http://trac.osgeo.org/mapguide/ticket/1999 ticket
1999 ) becasue I've discovered that more than 90% of tiles are blank and
useless, so more than 90% of time is totally wasted time.

I've just finished a last comparison test with several options on MgCooker
vs AIMS Studio.
Here you are my results.

The project is very light:
- a map with one base layer with 20 layers from shapfiles.
- the extention area is 50 km2, a very little town.

I use processor Intel Xeon X3440 @ 2.53Ghz 2.53Ghz (2 processor) on Windows
Server 2008 64bits.

Serverconfig.ini is configured as follows:
RenderOnly                         = 0
TileCachePath                      =C:\Program
TileColumnsPerFolder               = 500
TileRowsPerFolder                  = 500
DefaultTileSizeX                   = 1000
DefaultTileSizeY                   = 1000
ImageFormat                        = PNG

I've configured tiles for scales:
- 200
- 500
- 1'000
- 2'500
- 5'000
- 35'000

MgCooker performances:
- if I use standard configuration, 20 concurrent process with CPU 80%-90%:
1'012'588 tiles, 100 tiles each minute
- if I use official method with 1 meter pr. unit, 20 concurrent process with
CPU 80%-90%: 460'458 tiles, 100 tiles each minute
With MgCooker nothing changes if I redefine a closer bounding box

AIMS Studio 2012
- if I don't redefine a bounding box and with 10 concurrent process (the max
in AIMS Studio) I get 138'667 tiles to be generated
- if I redefine a bounding box closer to the town and with 10 concurrent
process I get 69'089 tiles to be generated

Both Mapguide and AIMS generated a big amount of blank and useless tiles,
but with AIMS I can reduce 460'458 tiles to 69'089 tiles.

Both products waste a lot of time in generating blank tiles (about 80% blank
tiles for AIMS and 90%-95% and more for MgCooker).

If somebody wants to take a try to the process to check and compare time,
I've attached this project to  http://trac.osgeo.org/mapguide/ticket/1999
ticket 1999 .

Are there anybody that has the same troubles? Any workaround?

View this message in context: http://osgeo-org.1560.n6.nabble.com/MgCooker-bad-performance-I-can-t-use-it-Help-tp4968806.html
Sent from the MapGuide Users mailing list archive at Nabble.com.

More information about the mapguide-users mailing list