I’ve been playing with Christopher Schmidt’s WPServer demonstration and I have to say it really makes me think about where GIS is going. Simply put, you can perform GIS analysis simply by clicking button on a simple web interface using all open source software. Jason Birch has a nice writeup on what you can do with the demo and I think most GIS analysts will be able to work though the tools without any instruction. But deep down we need to think about how this is changing GIS analysis.
This demo really can’t do much just yet, but the framework is there. I can imagine users using tools such as this WebProcessingServer and OpenLayers demo, rather than ArcView, Manifold or even QGIS to perform simple geoprocessing. Now this of course won’t replace the complex analysis that ArcGIS Desktop or even ArcGIS Server can accomplish, but why not use such lightweight tools for lightweight needs? I can’t stress enough the underlying message here.
Not only can you perform GIS analysis on the web, you can do it using freely available open source tools, you can do it with very simple websites, and it is very fast. This is definitely an area we all need to pay close attention to and I suspect BIG things are going to be happening here.
I had a talk with the Arizona State University MAS-GIS class last night and I wanted to demo this to them but unfortunately the demo was down. Hopefully some of them will be able to take a look at it today and see why this is so important to their future.
30 Comments
James:
Right on: basic geoprocessing over the web via an easy-to-use interface can satisfy a vast number of use cases of general users without hitting them over the head with a steep learning curve.
I talked to the guys at 52 North about their work with WPS and they passed along that they’re working on exposing more of the functions in GRASS as WPS. So the evolution towards more robust analytical capabilities may be closer than assumed.
BT
I’m a little more skeptical. I think web based GIS analysis is going to be about like web based word processing apps. That is to say, an interesting and usable solution that hardly anyone actually uses.
Of course, I could also be proven wrong about web based word processing apps…
Who wants a map on the internet anyway? I suspect that Google Maps will die because everyone wants their maps as hard copies.
It isn’t so much as IF, but HOW SOON. Even ESRI is pushing web based GIS analysis. If ESRI gets it, we should too.
@Lefty:
Do you really believe Google would be unable to come up with similar tools to Christopher Schmidt’s WPServer demonstration, which is excellent by the way, if they the desire?
You just have to look at their editable driving route tools to know that they’re headed in this direction already.
Hi James,
That was a great talk you gave our MAS-GIS class last night…and yes, you did a great job pointing out the importance of web processing/ portability and open source layer evolution. Thanks a million James…you are an excellent speaker. Bob
Thanks Bob! I did have a great time talking with folks last night. I was interesting to see what folks are learning in GIS these days.
I think Lefty was being sarcastic.
ESRI has pushed things in the past that didn’t pan out. Anyone still using ArcReader?
One of the reasons I’m skeptical – when ArcIMS came out, you saw online mapping services come out by the hundreds overnight. Where are all the ArcServer based geoprocessing services? They should be all over the place by now. I’m just having trouble visualizing a killer app that uses web based geoprocessing I guess.
@Chris C.
I don’t doubt Google will be offering such tools. I don’t think that reduces the importance of what we are seeing here, do you?
Those who don’t understand the paradigm shift we are witnessing here have their heads in the sand. Folks aren’t just thinking about this kind of functionality, they are actually doing it today.
Avoiding it will only hasten your obsolescence
@Ben Slater Just because everyone hasn’t switched their external IMS app over to Server with 19 spatial analyst tools inside doesn’t mean the Enterprise shops have. Large companies could be cutting costs on their desktop licenses by deploying Server GP tools to their users…. Just a thought… Who would you rather give a complex GP task to, a guy on the internet or your already trained internal user?
That’s a good point. Serving maps on the internet is great and was a no-brainer for most organizations. But I can see why people would keep the geoprocessing parts within an organization – which would explain why I don’t see them all over the internet.
It’s great to see these GP tools coming to the Internet. It proves the point that browser based tools can do more than simply display maps. However server centric GIS is about more than deploying standalone GP tools. Its about supporting spatially orientated business work flows that often rely on a spectrum of GI services from simple map display to complex GP models. Its about moving business logic from the desktop where its only accessible to GI professionals to the server where it can be accessed by the Chief Executive. Its about integrating GI into legacy line of business applications using IT protocols. Online mapping services may be used to present the results of these analysis but its likely that the “grunt work” will be done on the other side of the firewall.
Postgis is leading the way in ‘Open Source’ GIS analysis. Though most of the functionality is hidden from the user. Google Maps is great as a front end which non-gis users are familiar with. Combine the two have you have your ‘free gis’ with the geoprocessing. http://postgis.refractions.net/docs/ch06.html
ArcReader… Acrobat killed that. “thanks for the map but can i have int pdf format please!”.
Mapperz
One reason I think we don’t see a lot of public ArcGIS Server sites and almost no public geoprocessing services is the performance. It’s no secret that ArcGIS Server is a resource hog – that’s just how it is. I doubt that GP services based on open-source software are going to be too much more performant – running a weighted focal mean operation across a state takes time and resources regardless of who’s software is doing the processing!
I believe the problem is that the scale out price with ESRI is high enough that many organizations are not willing to spend the $$ to give access to Joe Public.
In stark contrast, open source solutions become more cost-effective the more you scale them out. Once you’ve written the core code – the only cost to scale out is the hardware.
So, in a weird way, I see open source actually being more popular in the large shops where scalability is needed, but the cost of an ESRI Enterprise License is out of reach. Conversely, I see ESRI being more attractive to the smaller shops where ease of authoring/configuration and “kitchen-sink” flexibility is important. Unfortunately this is also where the pricing pressure exerts itself.
It will be interesting to see how this sorts itself out over the next few years. Will the open source solutions get so easy to install and configure that Suzie @ the County will setup OpenLayers MapServer WPS PostGIS? Will ESRI change pricing models to make scaling out more viable? Only time will tell…
Dave
Regarding the ArcReader bad-mouthing… I actually use ArcReader fairly often. For instance, I just gave a workshop in which I wanted partcipants to be able to browse around some spatial data that I was demonstrating on screen. There was no way I was going to set up ArcGIS on everyone’s laptop, and no way I was going to set up some web-based mapping application just for this presentation (not to mention the fact that everyone would have had to have web access during the course of the presentation…)
So I just created a self-contained ArcReader project that I stored on a thumbdrive, and then quickly loaded it on everyone’s machine before the talk. When the talk was over, they could either delete the ArcReader project, or keep it.
I disagree with Ben Slater’s first comment. I think that the difference between word processing and GIS is the data.
When word processing I’m putting in all of the data and I want it as close and as fast as I can get it. When doing GIS I may be interacting with several large data sets. Certainly I can download the datasets and work with them locally, but for many tasks a web interface allows me to work with overlapping subsets from their servers. Granted, some performance may be lost but that’s ok in most cases.
Postgis is leading the way in ‘Open Source’ GIS analysis. Though most of the functionality is hidden from the user.
@Brian Timoney
Do you see a revenue stream for geoprocessing over the web?
p
GAPE – Google Apps Premier Edition http://www.idgconnect.com/software/google_apps_in_the_enterprise_a_promotion_enhancing_or_career_limiting_move_for_architects/index.html
Anybody care to elaborate on “lightweight needs”?
@Paul,
Not to answer for Brian, but I see revenue models based on the creation of information products, as opposed to generic “GP Services”.
This would require that the provider has some secret sauce – an algorithm, or dataset that only they have access to, which produces an output with measurable value. I could see this playing in a B-2-B scenario more than B-2-C, but who knows.
Something like a “personal risk map” for insurance companies. You’d take the home and work address, create a route between them, then overlay traffic accident probabilities, disease surfaces etc to provide another – very custom – risk rating that the insurance provider can rate the custom on.
The main thing is that you’d need to provide something of value to a client. You’d implement a GP system simply because pre-computing all possible requests would be impossible or impractical.
Dave
@Paul, @Dave:
There is huge potential for revenue coming from geoprocessing over the ‘Net, however with the following caveats–
1) Please don’t tell the customer it’s “geoprocessing”, “GIS” or anything that would just confuse the general user with jargon.
2) Ease of use is paramount. We think piggybacking off the user experience of consumer interfaces minimizes the learning curve for a broad audience. And the point isn’t to reproduce desktop GIS over the web, it’s to add in just enough analytic capability to add utility while preserving ease-of-use.
3) This General User audience I have in mind doesn’t have extensive analytical needs–buffers, intersections, distance calcs cover most of their use cases. That these can be done within a spatially-enabled database can go a long way towards scalability and performance by leaving Server middleware out of the picture altogether.
4) I believe the great sea change will come with SQL Server 2008 + the .NET developer community. I’ll bet anyone, dollars-to-donuts , that in 18 months the majority of buffers and intersections being done over the web will be done by .NET programmers outside the ESRI sphere with no formal GIS background. I’m not saying that will be a optimum state of affairs, but the Microsoft juggernaut seems intent on making spatial + visualization in VE a key selling point to corporate America (unlike Oracle, which if you google ‘Oracle Spatial demo’, doesn’t seem to return a lot of results with links to, well, live demos–third-party blog entries and white papers don’t get me overly excited in 2007…).
5) The next tier of geoprocessing where we see promise is the interpolation of large point data-sets. If I can send 10K point values to a server and get back an interpolated raster in < 15secs that automatically overlays in Google Earth, GMaps, VE, whatever, then I’m giving the General User tremendous value by quickly offering a visual summary that allows him or her to identify prominent trends and outliers without slogging through individual records. Again, as GIS guys, we want to immediately launch into the IDW/kriging/Voronoi methodology debates. The user wants simple + fast and is more interested in quik-’n-dirty summarization versus the subtle arguments for each methodology (as offensive as that might be to those of us who enjoy the subtle arguments…).
6) #5 is why I’m intrigued by WPS. As mentioned elsewhere, the guys at 52 North are working on ways to port GRASS functions into the WPS framework. Of course the devil is in the performance details, but the promise seems clear enough.
Finally Paul, if you can figure out how to leverage your work with Amazon EC and create a transaction-fee model for these types of services, I’d be happy to sign my clients up. What we’re seeing among our clients is not a frustration with proprietary licensing fees per se, but with the configuration/IT Management burdens that come with the licenses. Most would happily write a check every month for services that would relieve them of the sys admin burdens and free up their core GIS personnel to focus on critical tasks such as data quality, analysis, and cartographic display.
GIS people freed up to do GIS–a revolutionary concept.
Brian
A few things that concern me regarding online GIS:
1) Access to good base data (precise and accurate) 2) Control over that base data – appearance, attributes, etc. 3) Black boxing of geospatial knowledge
The gains in online mapping (graphics) have been phenomenal but it’s not GIS…yet.
Attribution and topology are the two big things that I don’t see mentioned much. Are they important anymore, have we just not gotten there yet, or are they hidden?
To me the strength of a GIS is it’s ability to be adaptable – in both analysis and representation of geographic data.
But it seems that things are moving forward and fast and they are interesting times to be participating in. (Dave said it better!)
@Dave
I agree with the secret sauce, and like your insurance example for online GP. It begs for a subscription based transaction service. I do not know much about the insurance industry. Do you think there is sufficient demand?
@Brian
Building a scalable GP solution on AWS, combined with a viable transaction/subscription service, would certainly appear to be a logical endeavor. I think it is a bit premature to be “signing up clients” but we could have a useful conversation.
@Paul,
I really don’t know – my point was more that whenever there is a B-2-B transaction, ppl will look to automate that. As we see geospatial awareness move deeper into the traditional IT “space”, I’m sure we’ll see more need to provide GP automation as a part of the solution. As others have said – it’s going to be called GP, but it’s just “what it takes” to provide the service.
I’m off to Mexico for a needed break – Cheers!
I’m an exploration geologist and currently use a web-based mapping tool provided by our local (State of Victoria, Australia) government. It’s great for quick analyses, lookups, and mud-maps of our mineral leases, combined with publicly available geological data, not to mention a LOT of historical data – data that we just wouldn’t have in-house, or maintain. I can create a map pretty quickly, and print to PDF or download shape files. We can drill through hundreds of data layers to access current land-holding information (including pastoral and industrial), previous exploration reports, borehole data with lithology and assay results in some cases, govt survey geological interpretations and mapping data, and do basic dataset queries (selecting by attribute, point, location etc). Pretty handy indeed.
I hope people are not arguing to move everything to the web and not have a “back-up” locally. The agency I work for, seems to be heading down that path. Put everything on the web and that’s all we need, bah.
What happens when access to the web goes down for a extended period of time? Have people sit around?
I don’t think people are advocating putting all their eggs into one basket. I haven’t seen someone yet bring into the debate the need to have two lines, one in-house and one on the web. Granted if more moves to the web, the less one would need in-house capabilities. I would think those in-house capabilities would never go away, just shrink.
KoS
I work in land conservation as GIS technician and I see web-GIS applications like WebProcessingServer as very useful for staff whose main duties are not GIS related. They tend to use GIS for very simple applications and may not have time for other than basic GIS training. I can say this because I observe that even as most staff have ArcGIS on their desktop, some prefer to use Google Earth because they can communicate/exchange data with people out of the organization (of course not all of them, and not all the time).
The main challenge in my opinion for webGIS is to be able to access accurate data online (parcels, recent images, etc.) and have a place where analysis results could be stored/accessed/linked to organization’s databases. More complex analysis could be done by GIS technicians/analysts using desktop applications.
Great blog,
There is an Application that wraps around Arcreader called GeoSync XG. Pretty interesting app….
http://www.mapsync.com/GeoSyncXG/index.htm
There is an Adobe Flash based framework that can do GIS analysis. It can be integrated with various back-end technologies like PostgreSQL+PostGIS. You can see a tutorial about creating a Web GIS application here:
http://www.flashnavigator.net/tutorials/web-gis-tutorial-wigle/
2 Trackbacks
[...] GIS Analysis for the masses (tags: OGC:WPS geoprocessing) [...]
[...] I’ve always thought so, and there’s a great discussion of this topic in this James Fee GIS Blog post. James shows an example of a web-based geoprocessing application, and surmises, as I do, that this [...]