I spent most of the afternoon talking with folks about web mapping systems and how to choose a direction to go. What is painfully obvious when you start laying out the different frameworks, APIs and servers is that there is just no clear answer as to what system to pick. When it comes down to it, what you are really after isn’t the system itself, but the product the system produces.
I think we need a taller pole
I talked a little about the different marketing direction I was taking last month, and this is partly it. It isn’t so much as abandoning one product for another as some had written in the comments, but more looking at the systems as a whole. Web services are web services, no matter if it is ESRI, Google, MapServer or even Manifold running the server. There is one constant that I’ve noticed in the past 2 years is that the server behind the product has very little to do with the quality of the product users see. You see users don’t care that it is Google, .NET, open source or even Manifold. They just want their experience to be useful.
Of course you may say it is easy to sit back and not worry about servers when you are in a IT structure that may dictate that you go one route or another. That is a huge reality in the for many implementors but it shouldn’t stop you from focusing on the end product and not on the server.
As I was on the airplane flying to New Orleans this morning I was thinking about where GIS application developers would be in the next 2 years.
Hosted Web Services
Now don’t start flaming me yet. I do realize that hosted web services aren’t possible for everyone or even wanted. But as this space grows beyond the traditional users of GIS, you’ll start seeing organizations wanting to get the power of maps with their products, but not have to worry about hardware, software or datasets. We’ve seen this with so many companies using Google Maps and Virtual Earth in basic web applications, but with the development of GeoJSON I think we are going to see much more integration of hosted web map services with existing business models.
Hosted web services is just the logical next step when you are not interested in the server itself, but the services that it provides. We’ve seen tons of innovation in the Amazon Web Services space so we probably need to keep one eye out there as well. I’m interested to see how this all plays out.
22 Comments
I can see that hosted web map services argument. I mean isn’t that what Google and Microsoft are doing with their APIs.
My new project for this year is to migrate our old ColdFusion CMS to SharePoint. I was planning to just embed VE right into the SharePoint and drop ArcIMS all together.
Kudos James! It’s time more people in the geo industry get past the platform hissy fits, and actually pay attention to what the users need!
The growth area for “geo” is not GIS Pros, or even their acolytes. Growth will come from expansion of “where” into traditional business workflows, and be used by the spatially challenged. Those organizations will not want to have all the geodata infrastructure, licensing etc, so hosted services are a great model. I believe that high-performance, focused applications that scale and have reasonable licensing models will dominate the emerging market. There may be a place for do-it-all behemoth software in the mix, but it’s going to be back-burner.
Change is coming fast and furious…
Dave
Having done an actual application or at least a techonology demonstration with all of the products listed above and some others as well in the last year or so, I completly see where someone looking at platforms for platforms sake could really get overwhelmed by the array of choices. I agree with James that it is the end result that matters and that is what one should focus on when trying to match up the platform for a client. I think this is one of the strengths of OpenLayers. It can act as a front end for nearly every server backend or major internet mapping platform out there. You could mashup data from ArcGIS Server, ArcIMS, WMS, WFS, GeoRSS, GeoJSON, Virtual Earth, and MapDotNet all in a single very Web 2.0 map with surprising speed. Another good example of a results oriented approach rather than a platform specific approach is Placebase.com and the Pushpin mapping platform. This is built on an amalgamation of numerous other technologies, data services, and some properitary code. They are using( among other things): ArcGIS, PostGIS, TileCache, kaMap, Proj4JS, Yahoo UI, and a very immpressive data catalog. They also offer RenderMap, a service in which they take your data and/or an ArcMap project (.MXD file) and cut custom tiles for you that you can have and use for yourself in any application you want or you can have them host for you. The results are a fast, nice looking, and HIGHLY customizable Web 2.0 style map. Someone using the map would have no idea what all was behind it all. They would just see a very familar, GMap – like interface. I think we see more & more of these type of services for both general public and the corparate world.
I don’t really consider metacarta a web mapping system in the same regard as ESRI or Google products. Metacarta is really just an API that can integrate on top of any of those.
Wow, what a post James. I need to start thinking about this shift to web services from my current way of thinking (ArcIMS). We are getting ready to put out a RFP on our system and I might have to rethink some of what I wrote. Thanks!
J Wallis: The graphic came from a presentation I gave. It is a little out of context, but I figured it would illustrate part of the issue I’m talking about here.
I try and mix in some non-Simpson’s graphics from time to time.
@Donald:
Have you seen this entry?: http://virtualearth4gov.spaces.live.com/Blog/cns!369B39F890CE30C1!623.entry
It’s interesting to know that there are others trying to bring mapping capabilities into Sharepoint. Not to get off topic or hijack this thread but we’re also looking to do the same thing along with tying into our existing AGS Server services.
To the actual topic:
Acquiring agreements w/ Microsoft/Google is something that is VERY appealing to the (my)large corporation. Being able to make a business case and demonstrate a business need for software is something that the “higher ups” need to hear whenever we (I) get “excited” about implementing new methodologies or strategies involving a GIS.
Couldn’t agree more with this post – well said.
@Donald: Be aware that if you bring Google Maps behind a corporate firewall you have to go with a pay-for-option (abiding by the current Google license) — and Sharepoint can integrate any web mapping solution; however a canned out-of-the-box web part for VE is a no brainer IMO — at a minimum it keeps MS customers from looking to other providers. This is a post from the Sharepoint Dev team on this topic.
Adam & Paul – thanks for the links. I’ve seen some web parts for VE, but I didn’t feel like paying a company for something that is “free”.
This is an excellent post and ought to be a must-read for anyone implementing web apps. It almost perfectly sums up where I feel I am as I try to decide which direction to go on a few projects. OpenLayers, Mapguide OS, and Mapbuilder all look like interesting possibilities right now. About the only thing I’m sure of is that I can’t get rid of ArcIMS soon enough, and also that I’m not sold on AGS as being the front end for our web apps.
I don’t know James, you seem to be contradicting your position with your 12/5/07 post “People just don’t get it”. It sounds like you see the difference between GIS and non-GIS. Where GIS is defined, appropriately, as an Information System (much like CIS) and is still ‘old world IS technology’ versus new GIS…which really isn’t literally GIS, as in an actual Information System that has processes and work flows, inputs, outputs, etc. but is more enabling technology for another set of users, but isn’t really a defined ‘Information System’.
The thing people also need to consider is if you really need a GIS, or just a map with points on it? Services like Google and Microsoft are good for giving up canned data or basic functions.
But when it comes to doing custom doing custom logic and functionality to do true online analysis I think you are still pushed towards the ESRI, MapDotNet, MapServer or OSGeo route.
I don’t know enough about MetaCarta to speak to its development potential as a spatial analysis tool.
@David Wright
True, but you can build all that lovely analytic functionality on top of an openlayers/VE/tile cache powered site. You’ll get a boat load more performance, have a lot more options in terms of UI (prototype & MS Ajax & name-your-ajax-library-de-jour) and save a ton on your licensing (assuming you’re using AGS for the analytics).
I be clear – these other options are not for Joe GIS at the county who needs to stage up a “map”. AGS is great for that – easy and who cares about performance b/c there aren’t many users. These options come into play where you have 100’s to 1000’s or more users of a focused application.
Dave
Dave
@Bouwman
Yah, we are not even on the AGS side yet, but pushing out a couple hundered layers of data to ArcIMS for viewing.
Moving to other platforms makes sense once we have a fluid method to go from Arc on the desktop to the internet and have the same level of control and access. I know many locales are already making this kind of move but in a long established shop where you have other infrastracture apps based on the ESRI platform you tend to be forced to maintain it.
GeoPrise is actually going very much the MapDotNet route, where our UI supports consuming data from a variety of sources, so we just provide you a unified publishing method with a consistent user experience with all of the buzzwords supporting it (AJAX/WCF/WPF etc) but we have a way to go.
Not really. What I’m saying here is if traditional GIS implementors want to reach out to clients they never would have looked at 5 years ago, they need to look at hosted services. Just because the local dog washing service doesn’t work with desktop/server GIS doesn’t mean that they can’t take advantage of GIS’ benefits.
Maybe the terminology was sloppy on my part, but the message is services will allow you to help them out.
If think if you look across most technologies in general the concept of desktop software or applications on servers are being replace with hosted services that can be reached via the web. Personal finance, email, tax preparation are just a few of the things that are shifting away from traditional desktop applications. As users/companies/governments feel more comfortable with security, you will see larger shift to offsetting cost by having entire applications hosted as web services. Why would you want to pay for the connection, machine, software and personal when you get the same uptime from a hosted service?
Indeed – well said. Not only do I agree with you, our client base and market is basically starting to show what you mention. Some clients (particularly small ones w/ no IT) are looking for hosted solutions. Others, regardless of hosted or not are looking for solutions period, and are either less concerned with what runs it (as long as it works) and/or are jadded by some of the long time vendors who’s prices are high and bang-for-buck questionable.
James, I was going to say this was the best post I’ve read all year, but given its January that doesn’t see appropriate. How about this is the best post I’ve read in the last year?
Thanks for thinking about this stuff. I really helps those of us who don’t have the time to brainstorm about where GIS might be in 2-5 years.
Keep up the great work!
For a moment, I thought James was explicitely plugging deCarta’s Hosted Web Services (HWS). As the product manager for HWS, I couldn’t agree more that for most users, it’s the service, and not the platform, that should be the focus. There are inherent limitations of web services, such as where the data resides, and in-house hosting of geospatial platforms is not going away, but for new participants in this field, and businesses where LBS is not core to their business, hosted web services should save them time and money!
Ok, I’ll play the roll of the old curmudgeon.
Yadda yadda yadda. We all know that the landscape of our industry is changing, and the old guard should be on notice, but show me the solution that seamlessly bridges that gap between the legacy GIS resources and the contemporary commercial services. There are some products that have promise but they are not the panacea they are made out to be. I’m as excited about all this as the next guy but with all the options the variables involved in integrating the growing number of innovations are staggering. I fear the kludge will soon be king. Whoever provides a stable product and or service to paying customers who can plug and play their way to consolidating these disparate resources will be the true winner. The GIS middleware market is fertile ground.
Hi James
I have just recently pondered the same things about web GIS. We have progressed over time towards a Mapserver based WMS with an Openlayers frontend. Why Openlayers..well it mimicks Google the best and that is what the end user knows…so there is no need to “train” them on using navigation tools etc. The other reason of course is price. And yes..end users do not care about what the backend is…as long as it looks good, is easy to use and does what they want.
You can see my train of thought at my company blog. I will probably get a flaming for my thinking…..but I welcome a constructive argument!
Cheers Andrew
Interesting thoughts James. When I first started thinking about this after reading your post, I was of the mind that maybe this isn’t something for me, but the more I think about it, the more I realize that I’m already transitioning toward this model outside of GIS. As someone else said, “It just makes sense”.
Thanks for the great discussion as always.