Hi,
Before you start optimizing anything you should evaluate if the performance that you have is normal. For example in this 11 years old benchmark https://www.idee.es/resources/presentaciones/JIDEE07/POWERPOINT_JIDEE2007/PowerPoint.7-Mapserver_vs._Geoserver.pdf which was made with a computer that is today classified as a toy
(Dual core (1.8Ghz per core). 2GB RAM. 7200RPM disk. Linux. PostgreSQL 8.2.4. PostGIS 1.2) shows much better performance.
Shapefile of size 6 MB is really very small dataset. If it takes a second to render it with a powerful computer there must be something special in the data or in your installation. I am sure that there are people on this list who get better results with Raspberry Pi.
Can you share your 6 MB dataset? Or could you repeat your tests with some public data for example from http://www.naturalearthdata.com/downloads/?
-Jukka Rahkonen-
···
Lähettäjä: Tripple Tee [mailto:trippletee96@…84…]
Lähetetty: 9. elokuuta 2018 7:51
Vastaanottaja: geoserver-users@lists.sourceforge.net
Aihe: Re: [Geoserver-users] CPU drain
Shape file is not doing well even on a powerful CPU. I have the same set of data on a proper server with Xeon chip (Virtual Windows 2016 with 10 vCPU) the performance is a mark improvement compare with the poor Celeron. The CPU usage jump to 15% every time I zoom in or pan, but it still take a second to render.
I did some googling on Java and multi core/thread CPU. Seem like Java and most software can not take full advantage of the multi core/thread provided by hardware.
Thanks Brad for the info about PostGIS, I might have to stuck with shape file and it lean, my design goal is portable:
1 geoserver on main server, when a team going away, all user need to do is to robocopy the whole folder from the main server to the portable server. Lucky GeoServer was written in Java so I can achieve this while using different OS (main server is Windows 2016, portable server is Linux)
Might be able to do this with PostGIS but I will have to copy the database from Windows to Linux, but it will not be a simple process for user.
Sent from Mail for Windows 10
From: bradh@…7296…
Sent: Thursday, 9 August 2018 11:43 AM
To: ‘Tripple Tee’; geoserver-users@lists.sourceforge.net
Subject: RE: [Geoserver-users] CPU drain
There are a bunch of tips at http://docs.geoserver.org/latest/en/user/production/index.html and you need to benchmark (e.g. with jmeter) against your expected workload when performance tuning.
In general, shapefile is not going to be very high performance, as noted at http://docs.geoserver.org/latest/en/user/production/data.html#use-a-spatial-database
Brad
From: Tripple Tee <trippletee96@…84…>
Sent: Thursday, 9 August 2018 11:16 AM
To: geoserver-users@lists.sourceforge.net
Subject: [Geoserver-users] CPU drain
Hi,
Please give some pointer on optimising for better resource consumption
I am running GeoServer on Linux with Intel Celeron 4-core processor, which I know is slow
When I open an individual layer (or layer group) of a region 200km each way (with details like roads, buildings …) the shape files size is under 0.5MB it loads quite comfortably, effortless to zoom and pan.
When I open a layer (or layer group) of the whole world with less detail (only land and water lines) it seem to struggle. The size of the shape file is around 5-7MB, I can see the CPU spike and it takes seconds to render as I zoom in or pan the map. Note: I am preview this layer on the default 750x400 resolution box. Should GeoServer only query data of the sub region for that preview box ?
1 - Is this the expected performance ?
2 – How can I optimize it to stop the CPU spike and lag,
Thanks
Sent from Mail for Windows 10