Why it is still not common to publish and use WPS?

Question

I tried to find some online platform-independent GIS services that could be called by any programming language and doing various GIS operations - like intersection, data format conversion, transformation and so on. WPS standard is a hot candidate for that (like in case of wps-grass-bridge) but unfortunately I cannot find such (or similar) services online - published by some provider. Most of WPS that is possible to find are just for testing purposes or are not working anymore (like from this post: Looking for a Public WPS Service).

I am aware that some operations may be very time and processor consuming but still there are many simple operations that could be managed and offered by this way - but they are not. Does anybody know why? There are many libraries that allows to publish WPS... Can be the problem in difficulty of using WPS for end users? Or people just don't want to use such operations online? Anyway it is a pity because it would offer a creation of various new services over existing applications (like Google Maps) without complicated programming and with possibility of services orchestration...

So my question is: Why it is still not common to publish and use WPS?

Answer 1

What's the incentive?

For WMS and WFS, it is dissemination of data. Even this has taken a long time and thanks to the various "Open Data" initiatives, they are a great way to do this. I am thankful several institutions are choosing to have more of these every day.

WCS, which is heavy in bandwidth for anything significant, is already a much higher cost. Serving up pretty pngs (like WMS) is much different than returning a 100MB rasters in terms of money. That is why you see less of this.

For WPS, it is even worse. First, consider the case of the trivial services (buffer, intersection, route, etc). Those have nothing to do with the organization. So basically they are dedicating machines to you, so you can buffer your data. The organization pays the bill, they do not disseminate their data, they don't collect data, they just let you use the servers for processing. Ehhh Why?

Now consider the non-trivial services. An organization decides to publish some aspect of their workflow as a WPS. There could be some value - but this is definitely a project that would take a considerate amount of resources by the organization (think more than one person). Perhaps they want to publish their flood analysis processing or some weather thing.

First, you have the fear that it can be abused, like a DoS attack, because by it's very nature it is resource heavy (it is non-trivial after all).

Secondly, the benefit is not clear to the decision makers. By having the WMS or WFS they are "open" already, right? So why take the increased risk (and cost) of WPS? It doesn't make sense.

Answer 2

The simple answer is cost - I can stand up a small web map server that is useful on an old PC that is lying around the office. For a WPS I actually have to write some code that does some process and I also have to find a big enough machine to run that process in a useful period of time. And it still probably won't run the process that you want :-)

Finally if you think I'm going to think I'm going to provide a powerful machine that you can run arbitrary code on then I have a bridge for sale too!

Answer 3

Actually there are couple of reasons :

WPS is fairly new. It takes some time for OGC standards to be established (implementations, popularity in terms of use) if everything goes fine (where I have some doubts that I am listing below).

• First problem is ambiguity about what WPS is doing really. Its name is already confusing. It is not tailored to specific GIS domain. Although I have not checked yet but new version tries to address that issue (at least I heard so). If this issue is not solved then of course people will question what is the difference between traditional web service and WPS. Its interface is too generic still.
• Practical issues about bandwidth and organizational boundaries (in terms of exposing data etc). Basically in most cases WPS should sit next to the WFS or WCS (in the same local area network let's say) because organizations most of the time does not let the data going out or even if they are keen on doing that it is not practical at all.
• I'm working for a project to develop spatial decision making system and there we want to use WPS at some point but I can tell you now with my experience on WFS already : there are lots of ambiguities in the OGC specs and there are some very important things are missing at the moment (like units and description of data attributes). So all those issues create interoperability problems and makes it difficult to develop spatial analysis on top of the OGC technology stack (or I should say OGC specs).

However still I am optimistic about it but it requires time and effort. Finally another thing : I think OGC specs should be written by a group of editors. Then it might be more consistent and less confusing.