In the past month I’ve had 4 people tell me that they are probably not going to be using ESRI Server (ArcGIS Server, ArcIMS, ArcSDE) on their projects unless their clients specifically demand it. I’m not talking about 4 people off the street here, but ESRI MVP types. When asked why they felt this way, they all said it was because they have invested so much in their web clients that they no longer needed to be tied to ESRI Server products and their added cost. One also said that there was nothing in the new ArcGIS Server 9.2 that they would be using anyway so it made no more sense to pay maintenance. Interesting that this is happening now, after ESRI is finally delivering clients that developers can use (ArcGIS Explorer, Web ADF). When I mentioned the Web ADF to one of these folks, she just laughed and said they already put thousands of dollars into their own web client and they didn’t want to walk away from it. I guess if the Web ADF had come out at lets say 9.1, many people might be already locked into the ESRI server stack. But the delay has caused people to go out on their own and in turn they have looked at open source solutions.
The other issue is that the added functionality of ArcGIS Server doesn’t give them any value. Some of the functions of AGS are impressive, but in the real world they have almost no applicability. As one person said to me, “ArcIMS really hasn’t changed in the past 3 years, so why should I pay maintenance on such a product? I’m sure the new wizard based map creation back end will be helpful to some, but it doesn’t add anything to my development work flows.” Another said that AGS is impressive, but speed makes it very difficult to use. “We looked at a combination of ArcGIS Server and ArcIMS to try and get a handle on speed issues, but the cost was so great that we just abandoned that route and went with Mapserver. We had some cool stuff going on but in the end we couldn’t justify the cost.”
So implementors of server GIS products are beginning to look elsewhere because of perceived value and speed right at
the time ESRI is releasing their most impressive server update in years. As one business partner told me, “Desktop is so impressive that I can’t ever see using something else, but in the server marketplace, the competition might have surpassed ESRI.” Simplicity, speed and freedom seem to be at the forefront of web GIS and ESRI might be losing out on that front. We’ll have to see what ArcGIS 9.2 does to that, but it would appear that release might be 1 year too late.
36 Comments
So wait, are you telling me ArcGIS Server 9.2 sucks?
Well I can’t say anything about ArcGIS Server 9.2 because I’m on the beta. We can’t disclose anything about ArcGIS Server, good or bad. I’d love to tell you what I think, but until 9.2 ships I’m keeping my mouth shut.
James:
Another way to think about the issue is what do people really want and need from a web experience and what should be best left to a desktop environment.
When we explain to clients that measuring distances, buffering, clipping, intersecting, etc., can all be done by a spatial database (e.g. Oracle, PostGIS) independent of GIS middleware, using off-the-shelf web scripting (PHP, ASP) rather than dealing with ArcObjects, there’s usually a pause then a skeptical “Really?”
Our response is somewhat along the lines of “Yes; really.”
In short, my hunch is that trying to replicate large portions of desktop functionality on the web isn’t as useful as taking a small handful of spatial operations and making them as easy to use and fast as possible.
Brian
This is also true for other GIS : we choose also MapServer, in our case instead of MapXtreme, the MapInfo solution, for two main reasons : - our servers are runing Linux and we wanted a fast and unix/linux native solution - the cartographic server we developed had to be uploaded to our 120 local agencies, so the cost impact of a proprietary solution was far too high
I think Esri is still the leader for Desktop software. With its ArcGIS and all its Extensions it is by fair an uncomparable product for GIS analysis, editing, cartography… But at the server side, I see impressive products like MapServer and PostGIS (not to talk about Oracle Spatial) that are at the same time powerfull as much as ArcIMS and ArcSDE (if not even more) and they are for FREE. That meaning that you can get rid of license and mantaining cost, having the opportunity to be paid more for more significant customization. If some guy would ask me to design an Enterprise GIS system I would not have any difficult to reccomend him such a scenario (ArcGis Desktop, MapServer, PostGIS). But Esri has an huge installed basis so we will continue for many many years to work with Esri backoffice software. Plus there are plenty of customers who don’t accept customized open source software, but want warrantied commerial solutions like Esri can offer.
I work for a US government agency that is getting ready to send out a RFP for GIS servers that has no ESRI requirement. A first for us. We still have tons of ESRI licenses, but we want to start thinking out of the box and we like the value that open source GIS would appear to give us.
Interesting how everyone seems to be agreeing with James here. Doesn’t sound good for ESRI if you ask me. I’m not feeling any vibe from 9.2, but I’ll be honest in that I’m not a beta tester and I only read what is on the internet.
ESRI should open up, this closed door mentality is going to kill them, especially on the server end.
I recently attended a rollout seminar that ESRI had for the ArcGIS 9.2 release. There was a videotaped speech from Jack were he several times apologized for the quality of the software and how ESRI listened to what their customers wanted. When the presenters started the stuck with the same themes how they killed lots of bugs and implemented features requested by their customers.
In 10 years of working with ESRI I never heard this kind of “we care for the customer “. After reading this post I think I have an idea why.
What do you mean MVPs? I didn’t know ESRI had an MVP program like Microsoft.
Actually they don’t (well they have that MVP thing for the forums, but I’m not talking about that), but I’d consider these folks most valued professionals given how much ESRI software they have implemented (i.e. sold).
I think that once the final 9.2 licensing is released, we’ll really see if/how the tide will change – a lot hangs in how this sorts it self out at the end.
Since I’m on the beta – I won’t elaborate until it ships, but it will be interesting.
Dave
Concerning desktop vs server open source, the touchstone will be connecting ESRI desktops with open source servers, and I can’t understand why the oss community hasn’t made this leap forward. Look at this forum: people are refering MapServer+PostGIS+ArcGIS. If I use ArcGIS I want to use PostGIS directly and seamlessly and I want to connect to MapServer too from my desktop app (just like with ArcIMS). This would really boost oss server products in esri shops. And compared to developing any of these products I can’t see how developing an extension format for ArcGIS can be that hard.
When my company started its GIS department, we bought ArcMap licenses because it’s the standard and also the best desktop solution out there. When we got big enough to need a spatial database instead of tons of shapefiles, we got ArcSDE because it ties in so well with ArcMap. If we ever get around to doing web based GIS, I’m sure we’ll choose a product based primarily on how well it ties in to our existing infrastructure. I don’t see the open source solutions doing this right now (or if they do, someone should write up a nice tutorial to show how it’s done).
Duarte, unluckly the ArcObjects model, based on which is build ArcMap, does not offer us a PostGIS WorkspaceFactory (like for Coverage, Shapefile, SDE, Access GDB), neither the possibility to extend the abstract WordspaceFactory class for PostGIS. At the moment it is only possible to edit PostGIS data in ArcMap in a disconnected editing fashion, something like with ArcSde disconnected editing. To do so you can take a look at PgArc (Open Source) or you can implement a custom solution with ArcObjects, it is not so hard. To analyze data I think you can use the ArcMapInteropAddon, but I don’t know how it works because I have not this extension. With ArcMapInteropAddon you could view and analyze data published as WMS, WFS, or GML ie with MapServer.
IMHO Esri will never give the possibility to use its Desktop environment with Open Source served data.
I love the competition, which is only going to make all products better in the long run. In fact, I’m setting up a race between ArcGIS Server 9.2 and MapGuide shortly. I’ve got two different projects that I’ll compare the two on. Let the best technology win!
On a sidenote, I’ve heard that one of the key ArcGIS Server developers claimed ESRI made the same mistakes with AGS as with ArcIMS. Could be heresay, but I have a reliable source.
Does anybody have experience with Orbit FlashMap? I have a non-profit my company wants me to do some free work for and I’m thinking about using it for serving maps.
James, wonderful observation. I’ve thought about dropping our ArcIMS maintenance because we just don’t use it that much anymore. We’ve gone the KML/Google Maps route and our lives are much easier.
What got me about all this is no one really disagreed with you. I think you hit the nail on the head with this one. There is a shift on the server side and you have to wonder if ESRI knows it is here.
Paolo,
I wasn’t aware of that limitation. But as you say, there are several workarounds. I’m a manager not a developer, and as such I would evaluate oss server technologies very differently according to the ability to integrate with the whole GIS platform – and the desktop plays a huge part.
Even if you know the custom format would not be “native” – as long as it works it would be great. You could have a plugin datasource for reading and displaying. For editing you could use a under-the-hood chekin/checkout scheam. Maybe the future will be Transactional WFS. Could this be “the” universal editable format?
Duarte
James – your post sums up our thoughts on this subject quite well. ESRI owns the desktop and I don’t think that will change anytime soon. The server side is a different story. Give me Mapserver and Oracle Spatial over anything ESRI has to offer. This combo is stable and robust. The problem with ESRi on the server side is ewverything is way too expensive, and they seem to radically chage the way everything works every few years or so…
OK, I will be the first one to disagree. I think AGS will contain one very important feature, that will be make it valuable to lots of clients and I do not know any other software package permitting it: Publishing Geoprocessing Models to a Server with One Click!
ESRI has developed over the last 3 years a very very powerful geoprocessing environment allowing to combine Core tools, Python Scripts and visually set up Geoprocessing Models in a hierarchichal way and they even permit an integration with the most powerful Spatial ETL tool in the GIS Worls, which is FME (aka Data Interoperability Extension).
Now, you will be able to select a geoprossing Toolbox and publish it to the server so that everyone (and I really mean everyone being able to download the free AGX) will have the opportunity to use this Geoprocessing functionnality, if you make it accessible to everyone. This could be the calculation of routes, data conversion services and much much more.
And, honestly, I do not know any other Software Package, may it be proprietary or OSS, allowing you this kind of easy deploiment of serverside geoprocessing…
Sot AGS might have too much overhead for webmapping or even for some web-editing tasks, but for geoprocessing, I really think it is very innovative and this will, in my eyes, be the main arguments for selling AGS licences…
For alternatives to ArcGIS Desktop, there is a UK company which is fully OGC compliant with support of read/write for PostGIS (and more) with license packages similar to ESRI Desktop offered on the market today. The GIS is “Cadcorp SIS”, I never used it, but it seems interesting for PostGIS/Map Server opprtunity of development (on Windows of course). Also, JUMP and parallel solutions seems promising. They have done a presentation that can be seen on the web site of FOSS workshop in Switzerland on how OpenJUMP as Desktop can be used in a wide/enterprise SDI situation with PostGIS.
Hmm, guess for the IMS part this is true. At least for us as an international organization we always preferred using OpenSource like Mapserver and PostGIS, although ESRI products were almost free for us. But ArcIMS was always slow and heavy…
On the other side we are now thinking about putting all of our data into a database and accessing it via a single interface – and the impression we have is that there is hardly any way around ArcSDE. Or am I mistaken there? We use ArcGIS a lot for DesktopGIS, as well as ArcView and ERDAS…
>I can’t see how developing an extension format for ArcGIS can be that hard.
might be possible with 9.2, since esri now uses GDAL: http://tinyurl.com/gwac9 “ArcGIS 9.2 has adopted the Geospatial Data Abstract Library (GDAL) and supports adding custom formats through the GDAL format driver.” and even links to the gdal driver tutorial.
no word about using OGR for vector access tho’.
Jonathan,
I agree with your enthusiasm for the concept of distributed Geoprocessing Models, but anyone who attended the ESRI conference a few months ago can attest, at least in my opinion, that it will be a while before the technology reaches maturity. In nearly every example I saw of this, it was LAN-only and very simple analysis (a single point buffer, followed by a selection, and an attribute report of selected features, for example).
The overall idea is exciting, but I would caution one from consuming too much of the ole ESRI cool aid just yet. If they can drop the price of AGS by 75% and increase speed by 3 or 4 fold, then we’ll be talking. Oh and reliability I’m sure won’t be perfect either this first release, so it’ll probably be AGS 9.3 before the kinks are worked out.
sounds like you are talking to all the ticked off ESRI BPs whose map apis because obsolete/redundant with AGS 9.2. I’ve had one of those BPs trying to tie me into their proprietary API and I just didnt bite.
all in all the shift you worry about would only go to no-cost software. the open-source/free software paradigm doesn’t play a role in that decision. ogc-support may be another reason but doesn’t have to.
Related to AGS and it’s distributed geoprocessing models….i really think that OS GIS is already there and doing it wery well -> http://bnhelp.cz/mapserv/wpsdemo I would really like to see this speed on AGS + ArcIMS. And i believe that in the moment when GRASS GIS will have complete python bindings more and more people are going to use it. Why?…it is better and is free I just wish good luck to Jack but the times are changing…
sorry correct link is http://www.bnhelp.cz/mapserv/wpsdemo
Paul,
Regarding your remark that ESRI radically change direction every few years…. they are a software company and I think most s/ware technology shifts every 3-5 years regardless of the specific app. This does cause problems with backwards compat. etc. but overall I think moving a technology base forward is the correct thing to do… now the question would be has ESRI moved in the right direction? it would seem reading this blog no -
It will be interesting to see if ESRI moves towards Open Source, but I imagine they will wait and see how Autodesk’s venture pans out.
Interesting thread for sure
I was wrong to say in my previous comment that there is not a direct way to let ArcMap direct read/edit PostGis. After deep investigating i realized that there can be 3 approaches for reading custom data: 1) Custom Layers 2) Plug in data sources 3) OGIS compliant OLE DB providers
With custom layers edit is possible but you have to implement a custom edit tool. With Plug in data source you can only read custom data, but it has the nicest integration with ArcCatalog With OGIS OLE DB read/write is possible, but could be actually implemented only with ATL c++
dear:
i am majid mousavi.student of watershed management in iran. your information about gis was very interesting for me and rise alote question about gis. please help me . 1):what does backend means in gis 2):what does frontend means in gis
sincerely yours
Backend would mean the servers. For ESRI that would mean; ArcSDE, ArcIMS and ArcGIS Server. The front end would be the clients; ArcMap/ArcCatalog, ArcReader, ArcGIS Explorer, ArcExplorer, ArcPad and some would say web based front ends too such as the Web ADF or the classic HTML viewer for ArcIMS.
Perhaps I can offer a bit of insight, the obvious or a perspective to remind those professionals who posted above. I’d like to share a “fresh” opinion or prediction on the ESRI/Open Source “phenomena” taking place. I’m a GIS undergrad student and my generation knows nothing but ever-changng technology and for the same reason Microsoft made great strides for PC’s, ESRI entered a market with very similar demands. You’ll understand this and much more then I do, GIS that is but after semesters’ of ESRI training, I still project my shapefiles in Adobe on a Macintosh and Linux will most likely provide faster accessability for satellites and open source. ESRI to me is remarkable software but it might be living out its legacy.
We’ve been delivering web based GIS apps with the support of ESRI ArcGIS family suite. But now for the next project we dont want our client to be put in so much cost of ESRI. We want to deliver him cost effective solution and one which is compitible with .net. His project doesnt require much of the functionalities that ESRI provide. So, we’re looking for the options available except ESRI. Where we can look for?
What do u suggest? Your suggestion will be worth a million. It will be great if anybody can provide the comparitive details of options available OTHER THAN ESRI.
Thanks a Lot,
Very interesting and informative comments. Thank you.
Does anyone know of a good open source option to ArcGIS Image Server?
Rie, I’m going to suggest that the answer depends on the complexity of your needs. I’m going to Assume that no Open Source product is going to match Image Server’s performance, but I’ve used MapServer to serve MrSid aerials on a non-production basis and non-aerial tiffs on an in-house production basis.
6 Trackbacks
Short news: Ornithology revolution; early SkylineGlobe reviews…
The miniaturization of GPS tracking devices is leading to a revolution in ornithology, reports the Boston Globe, as smaller and smaller birds can be tracked.Yet another story about how Google Earth is empowering the individual, and yet again it’s set….
[...] James Fee’s report of a shift from ESRI to Open Source for GIS server technology has stirred up some interesting discussion. [...]
[...] Personally, I can’t decide if the big guys are squaring up for an opensource fight, or if we’re just seeing much more involvement in the opensource movement. Following on from the recent discussion about the use of opensource alternatives to ESRI’s main product lines, and further back, ESRI’s decision to include support for PostgreSQL as a database, hopefully this is all a sign of a gathering momentum. At first I thought this 52 North initiative was a bad thing- a dilution of focus, but maybe it’s just a sign that the snowball’s getting that bit bigger. More products, and some healthy competition has to be good, right? [...]
The Blog…
I’ve been reading about the ArcGIS 9.2 release and some of the new features with puzzlement. The only need for our group is faster servers, not more features. There seems to be nothing in the new release that will address our needs so what to d…
[...] I’ve been getting more and more .NET developers talking to me about my MapGuide Open Source move over the holiday. Seems like many ArcIMS developers are downloading and checking out MapGuide, trying to get away from the ArcGIS Server/Desktop conflict (they have to develop on 9.1, but their IT department wants to roll out 9.2). I’ve said this before, but every day I get more and more ESRI developers asking me for directions on how augment their toolbox with open source. Personally I’m beginning to see the power behind hit and the .NET integration is welcome. After my initial battle with PHP, I’m running very well and at least on this laptop MapGuide Open Source and AJAX Tile viewer. It seems to run faster than ArcIMS 9.2 and the Web ADF on my laptop, but I’ll be honest that isn’t a scientific observation. [...]
[...] me just update with a link to this article from last year “Are we beginning to see a shift away from ESRI Server backend to Open Source [...]