[mapserver-users] Display performance

[Lime, Steve D (DNR)]

If you have postgis already sitting around then it's pretty quick just to try it and see.  I'm assuming you're talking 20sec for everything, all 170K features? A spatial index doesn't buy you anything in the 'all' case. That is a lot of features to display at once and I wonder if one can even make sense of that much detail. A common technique would be display reduced or somehow generalized versions of the data at that level and then hit the detailed data as a user zooms in. Without knowing the data how you'd do that would vary. For example, you could actually rasterize the data and display the pre-rendered raster version small scales. Or you might be able to thin the data a bit. Lots of options.

Steve

From: mapserver-users-bounces at lists.osgeo.org [mailto:mapserver-users-bounces at lists.osgeo.org] On Behalf Of Chris Jackson
Sent: Thursday, September 02, 2010 3:29 PM
To: P Kishor
Cc: mapserver-users at lists.osgeo.org
Subject: Re: [mapserver-users] Display performance

Hi

20 secs is from turning the layer on in the interface.  The data is a model grid as polygons, the file size is about 40mb.  I agree 20secs is too long - I am a web developer in general and know users are fickle!  The other layers load v.quickly so happy in general with performance of smaller model grids.

The same layer in an old arcgis server app loads a few secs quicker (from SDE rather than file) and we just accepted it was just 'one of those things' and deadlines got in the way - mmm...

The app is living on a reasonable specced server living in a data centre and the old web app has been serving the offshore renewables industry fine for a couple of years or so, just have a need to move to an open source solution.  I was trying to get a feel whether moving to PostGIS rather than local file will speed things up - I get the feeling it is marginal.

Maybe I am missing another processing trick (admittedly I have done minimal work bar a spatial index).

Thanks for the swift response.
Chris
On 2 September 2010 21:07, P Kishor <punk.kish at gmail.com<mailto:punk.kish at gmail.com>> wrote:
On Thu, Sep 2, 2010 at 3:03 PM, Chris Jackson <webturtles at gmail.com<mailto:webturtles at gmail.com>> wrote:
> Steve
>
> That saves some work, but how long would you expect one that size to load?
> I have just realised (admittedly not looked at the data for a couple of
> years) that it is actually 170,000 records - takes about 20 secs to load -
> sound reasonable?

What does "to load" mean? Is 20 seconds the time elapsed from when the
user clicks on the browser to when the user sees the resulting image?

You are ripping through 170K features and creating an image. 20
seconds seems like a lot. How big and complicated are those features?
How big is the shapefile, in physical size (MB, KB).

What kind of computer are you using?

What other processes are running?

So many variables...

in any case, if you have a web app, and you want your users to remain
interested in it, you probably will want to bring down the time "to
load" to sub-second, or, at the most a few seconds. You will also want
to consider multiple users -- acceptable performance with one user
might translate into horrendous performance with several concurrent
users.




>
> Chris
>
> On 2 September 2010 20:50, Lime, Steve D (DNR) <Steve.Lime at state.mn.us<mailto:Steve.Lime at state.mn.us>>
> wrote:
>>
>> That's not that big a dataset. Probably faster from a local shapefile.
>>
>>
>>
>> Steve
>>
>>
>>
>> From: mapserver-users-bounces at lists.osgeo.org<mailto:mapserver-users-bounces at lists.osgeo.org>
>> [mailto:mapserver-users-bounces at lists.osgeo.org<mailto:mapserver-users-bounces at lists.osgeo.org>] On Behalf Of Chris Jackson
>> Sent: Thursday, September 02, 2010 2:29 PM
>> To: mapserver-users at lists.osgeo.org<mailto:mapserver-users at lists.osgeo.org>
>> Subject: [mapserver-users] Display performance
>>
>>
>>
>> Hi
>>
>> I wondered if anyone knew if displaying a 50,000 polygon dataset would be
>> significantly faster if called from a database rather than a spatially
>> indexed shapefile, or is it a similar overhead for both.
>> Thanks
>> Chris
>
> _______________________________________________
> mapserver-users mailing list
> mapserver-users at lists.osgeo.org<mailto:mapserver-users at lists.osgeo.org>
> http://lists.osgeo.org/mailman/listinfo/mapserver-users
>
>



--
Puneet Kishor http://www.punkish.org
Carbon Model http://carbonmodel.org
Charter Member, Open Source Geospatial Foundation http://www.osgeo.org
Science Commons Fellow, http://sciencecommons.org/about/whoweare/kishor
Nelson Institute, UW-Madison http://www.nelson.wisc.edu
-----------------------------------------------------------------------
Assertions are politics; backing up assertions with evidence is science
=======================================================================

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/mapserver-users/attachments/20100902/60739220/attachment.htm>