[mapguide-users] Differences between UMN MapServer & MapGuide...

Tue Feb 12 13:20:38 EST 2008

Hi Tim,

We've got experience with Chameleon, Fusion, ka-Map, OpenLayers and  
MapBuilder around MapServer.

Cheers

Paul

On 12-Feb-08, at 9:13 AM, Nolte, Tim wrote:

> Paul,
>
> You've finally helped me to figure out what might be going on with my
> setup. The thought has crossed my mind several times about moving to a
> different application framework. Your description of the Chameleon
> framework seems right on par with my experiences with it. What
> frameworks do you have experience using along with MapServer? Thanks.
>
> - Tim
>
> ----
> Timothy J Nolte - tnolte at ilpcs.com
> Associate Network Planning Engineer
>
> iPCS Wireless, Inc.
> 4717 Broadmoor Ave, Suite G
> Kentwood, MI 49512
>
> Office: 616-656-5163
> PCS:    616-706-2438
> Fax:    616-554-6484
> Web: www.ipcswirelessinc.com
>
> -----Original Message-----
> From: mapguide-users-bounces at lists.osgeo.org
> [mailto:mapguide-users-bounces at lists.osgeo.org] On Behalf Of Paul
> Spencer
> Sent: Monday, February 11, 2008 8:26 PM
> To: MapGuide Users Mail List
> Subject: Re: [mapguide-users] Differences between UMN MapServer &
> MapGuide...
>
> Martin, I disagree with your statement that MapGuide will have far
> more development and support going forward.
>
> Caveat Emptor: The arguments that follow are purely my opinion and are
> not based in empirical evidence, mostly just my observations having
> worked with both platforms.  Take them with a grain of salt :)
>
> MapServer has a large and active developer community spread across
> individuals and companies that actively improve the software on a
> daily basis.  The MapServer mailing list is large and very helpful,
> and there is a wealth of user-contributed documentation that grows
> with every release.  MapServer's code base is (in comparison) small,
> light weight, understandable.  It has a relatively low cost of entry
> for new developers wanting to do stuff.
>
> MapGuide, on the other hand, is still largely dependent on a single
> (but very large) benefactor and has yet to attract additional (core)
> developers outside of the core team from ADSK.  It does have a growing
> community support infrastructure through this mailing list, which is
> probably the first step towards growing the development community.
> The code base is quite large, complex, and not readily understandable
> by a developer wanting to get started with it.  None of this points to
> 'far more' development or support going forward.
>
> I would also argue that MapServer, when properly configured, is faster
> than MapGuide - at rendering maps.  If you need to generate map images
> for the web, MapServer is generally going to be faster than MapGuide.
> There are some fundamental differences, though, that need to be
> understood when making this comparison.
>
> 1) Enterprise scalability.  It has been said that MapGuide is intended
> for Enterprise deployment (it supports multiple 'web tier'
> installations accessing multiple site/support servers potentially
> through a corporate firewall and distributed across multiple
> machines. ) with the implication that MapServer is not Enterprise
> ready.  In fact, this is not true and it is equally as possible to set
> up a similar architecture using MapServer - the main difference is
> that MapGuide has an installer that sets this up whereas the MapServer
> solution would have to be designed by a knowledgeable group.  When
> requesting a map image, then, MapGuide has a minimum of three
> processes that need to be involved - the web server, the 'web tier',
> and the 'server tier'.  MapServer only has two, the web server and the
> MapServer process.
>
> 2) Statefulness.  MapGuide is stateful in that you have to establish a
> session with the server and refer to that session in every request to
> the server.  MapServer is stateless and has no inherent concept of
> sessions.  For applications that require statefulness, MapGuide is
> typically simpler to use whereas MapServer requires an external
> framework to make it stateful and that framework adds overhead that
> makes the performance of MapGuide perhaps better than MapServer.  For
> applications that don't require statefulness (such as serving WMS or
> Tile requests), MapServer is simply faster because it doesn't have the
> overhead of statefulness built into its architecture (granted that for
> WMS and Tiles - MGOS 2.0 only - you can request images directly
> without needing a session, so that is less of a difference).
>
> 3) Optimization.  Optimizing for MapGuide and MapServer are very
> different tasks.  For both, you really need to understand the
> expensive parts of drawing a map - is it accessing the data, filtering
> it, classifying it, or rendering it - and how to minimize bottlenecks
> in each of the parts.  I would argue that it is more feasible to tune
> MapServer's performance than MapGuide's.  This is, in part, because
> you have more options with MapServer - different ways of accomplishing
> the same task.  MapGuide is probably more tuned to start with but my
> experience is that beyond optimizing your data, there isn't much more
> you can do to tune MapGuide's performance by playing with cartographic
> options.
>
> For data access, MapGuide has FDO, OGR and GDAL support.  MapServer
> supports roughly the same set of data sources, except it doesn't
> support FDO and hence can't access SDF files.  DWF support is not part
> of FDO, and its actually more of an *output* format than an *input*
> format.  It also requires an activex control to render it, which makes
> it more-or-less useless to me and many others that need applications
> that support multiple web browsers and platforms.  An FDO connection
> type for MapServer has been discussed and seems likely to happen at
> some point, although no specific plans have been made - so MapServer
> will gain access to SDF at that point.
>
> Regarding speed of data access, MapServer is generally considered
> fastest when used with properly prepared shapefiles, although PostGIS
> is arguably on par with it.  MapGuide is generally considered fastest
> when used with SDF.  I have had very poor success with SDE access
> (MapServer is about 3-6 times faster at drawing from an SDE data
> source) in MapGuide.  I have no knowledge of Oracle Spatial in either
> platform, although MapGuide is supposed to be quite good with Oracle.
>
> Cartographically, the two are quite similar.  MapServer has quite a
> few more options for styling data at this time, but MapGuide has some
> really nice styling features that MapServer doesn't have.  Again, it
> depends on what your critical factors are.  I would say that in
> general it is possible to achieve the same effect with both servers
> unless you have somewhat specialized needs.  MapGuide is arguably
> poorer with raster data sources.  MapServer can do things like raster
> classification and raster querying, whereas MapGuide can just render
> them as-is (I think I saw hill shading show up in MGOS 2.0, which
> MapServer doesn't support).  On the other hand, since MGOS 1.2 there
> are some extremely complex cartographic things you can do especially
> with line work that are not possible at this time in MapServer.
>
> Coming back to Tim's original question, though, I think that perceived
> performance in this case is likely 'application' performance.  As
> noted above under statefulness, application performance doesn't
> compare directly because you have to use another framework with
> MapServer.  In this case, Chameleon was built on pre-ajax web
> technologies and is particularly poor compared to MapGuide, Fusion,
> MapFish, MapBuilder, etc which have all kept up with, or been written
> for, the times.  Chameleon is quite horrible in fact - it does all its
> calculations on the server, creating high load and reducing
> responsiveness of the server under multiple simultaneous user load,
> and it reloads the page on every request which multiplies the
> problem.  'Modern' frameworks put a lot of the processing on the
> client side (in javascript), which reduces server load dramatically
> and allows a server to 'scale up' to meet higher load, minimizes the
> amount of information that needs to be transfered back and forth, and
> uses ajax techniques to keep the application responsive even while it
> is waiting for server-side processing to happen.
>
> So at the end of the day, my feeling is that the problem is with the
> framework that Tim is using around MapServer, not with MapServer
> itself.  The real question is which framework should Tim move to.  And
> on that particular subject, I don't have enough experience with
> platforms other than Fusion to comment on effectively.
>
> Cheers
>
> Paul
>
> On 8-Feb-08, at 4:28 PM, Martin Morrison wrote:
>
>> I don't have statistics, but years ago I played around with  
>> MapServer,
>> it was all right, but was not polished or speedy.  I find that
>> MapGuide
>> can be both.  MapGuide also will allow for more types of data SDF,  
>> DWF
>> to name two.
>>
>> I think MapGuide will have far more development and support going
>> forward.
>>
>> Just my 2 cents.
>>
>> Martin
>>
>>
>> -----Original Message-----
>> From: mapguide-users-bounces at lists.osgeo.org
>> [mailto:mapguide-users-bounces at lists.osgeo.org] On Behalf Of Nolte,
>> Tim
>> Sent: Friday, February 08, 2008 4:24 PM
>> To: MapGuide Users Mail List
>> Subject: RE: [mapguide-users] Differences between UMN MapServer &
>> MapGuide...
>>
>> I guess what I am looking for is what are the strengths and  
>> weaknesses
>> of both. Basically our MapServer currently serves both Shapefiles (39
>> layers) & Oracle Spatial (11 layers) data. My biggest concern is
>> user-perceived performance. Our current server config is a 3.06GHz
>> Opteron w/ 8GB of RAM. I've tried to tune our data as best I can
>> however
>> I'm still concerned about the end-user perceived performance. As most
>> people are, I'm up against people expecting our MapServer to perform
>> with the same speed as Google Maps. I'm also looking for something
>> with
>> a very visually polished development framework. I know that looks
>> aren't
>> everything but it does offer some credibility to the information  
>> being
>> presented. Thanks for any input anyone has.
>>
>> - Tim
>>
>> ----
>> Timothy J Nolte - tnolte at ilpcs.com
>> Associate Network Planning Engineer
>>
>> iPCS Wireless, Inc.
>> 4717 Broadmoor Ave, Suite G
>> Kentwood, MI 49512
>>
>> Office: 616-656-5163
>> PCS:    616-706-2438
>> Fax:    616-554-6484
>> Web: www.ipcswirelessinc.com
>>
>> -----Original Message-----
>> From: mapguide-users-bounces at lists.osgeo.org
>> [mailto:mapguide-users-bounces at lists.osgeo.org] On Behalf Of Jason
>> Birch
>> Sent: Friday, February 08, 2008 4:05 PM
>> To: MapGuide Users Mail List
>> Subject: RE: [mapguide-users] Differences between UMN MapServer &
>> MapGuide...
>>
>> I guess I should ask what your reasons for considering moving away
>> from
>> MapServer are.  What needs do you have that it is not meeting?
>>
>> They both certainly have their strengths and weaknesses, but I don't
>> know if any recent in-depth analysis of the differences.
>>
>> Jason
>> _______________________________________________
>> mapguide-users mailing list
>> mapguide-users at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/mapguide-users
>>
>> _______________________________________________
>> mapguide-users mailing list
>> mapguide-users at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/mapguide-users
>
> _______________________________________________
> mapguide-users mailing list
> mapguide-users at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapguide-users
>