The Open Source Debate…
Posted by Paul Vincent
There was an interesting discussion at last weeks International Summer School on the merits or otherwise of academia going the open source route for their software research. I was reminded by this by Seth Grime’s recent post on the topic of open source. The ISS discussion followed these lines (per my recollection):
- I was warning the academic community about ignoring the prior art from commercial vendors and avoiding work they had already commercialised (whether open source or not) - in other words, not reinventing any wheels.
- Google’s John Reumann then countered with the obvious benefits to researchers in extending open source projects rather than utilizing closed source products (i.e. access to extensible code, learning from the algorithm implementations and approaches used, etc).
- I suggested that for a research project, spending time learning how a product was coded (i.e. delving into the source code) might not necessarily meet the aims of the project (especially if the research is simply exploiting the relevant software, rather than altering it etc).
[Indeed I made the observation that much open source software is used because of its "free to download" nature rather than for the ability of, for example, business system developers to modify the core source code to their needs.]
TIBCO like many software vendors deals with combinations of free-to-download, open source, and closed-source-paid-for software. Various examples across the board are:
- TIBCO General Interface is an open source AJAX toolkit (used on a few CEP projects for custom dashboards, for example)
- TIBCO BusinessStudio is a BPM development environment with a free-to-download community edition
- TIBCO BusinessEvents is a your typical high-value paid-for software
- TIBCOmmunity is the free-access developer Q&A resource.
Of course, even the non-open-source products utilize some open source projects: BusinessStudio is based on / extends the Eclipse IDE project, for example. They also utilize (/embed) other commercial products - for example TIBCO BusinessEvents embeds (and extends) a distributed cache technology from a third party.
For researchers this is a tough decision, as some open source projects are simply commercially-developed products that were not competitive enough to be sold, or community projects targeting more general software. The main benefit to academia is obviously being free-to-extend, at a risk of being commercially-backward. Vendors like TIBCO have academic programs to try and encourage research usage, but this is still “betting on a particular horse” (albeit a good one 
).
So the overall advice? Academic researchers should not be shy about asking for evaluation versions of commercial products before judging what to build their research on…
Tags: open source
Complex Event Processing (CEP) | 26 August 2009 |
5 Comments
Other Links to this Post
RSS feed for comments on this post. TrackBack URI
By Glenn, August 27, 2009 @ 09:43
I think you’ll find that the “many eyes” collective intelligent nature of the more community oriented open source projects leads to layered architectures that are easier to study and understand. See http://www.dynamicalsoftware.com/ossCorpSite.pdf for more elaboration on that. For example, Esper’s events, patterns, and plug-ins make it easier to extend and customize for your desired CEP behavior.
By Paul Vincent, August 28, 2009 @ 06:44
Glenn - probably you are right. The question is how “community” oriented are many “open source” projects. Probably we need some better categorization for the term: source available (eg JESS), source freely available, source available and community coded, etc.
Cheers
PS: of course, closed source software is also often architected in layers to facilitate new developments more easily
By Hans, August 29, 2009 @ 13:31
Hi Paul, regardless of whether you offer downloads, there is another issue: free access to documentation and information about the product.
I know that TIBCO looks on such free information as a net loss. It just serves for comparison shopping, copying features into OSS and product evaluation that is unguided by the sales team.
But in the case of CEP, it’s also a big part of why there is no CEP community to speak of. No one can blog about the CEP landscape, because there is no public information on half of the products.
So to the extent that you would like more of a CEP blogging community (and I don’t know how much you care about that, no criticism intended if you don’t care), you would want a little more disclosure of information on BE.
By Hans, August 29, 2009 @ 13:33
Sorry, my previous comment was meant for your post on Forrester/evaluation downloads. I was reading a couple of posts on the site.
By Paul Vincent, August 30, 2009 @ 11:07
Hi Hans - regardless of what you were replying to it, is a fair comment.
- downloads of products or docs
- open source product / community feedback on product (eg TIBCOmmunity)
Once we move to Eclipse we can consider a Community edition of BE, much like we do for BPM with BusinessStudio (ie freeware BPMN editor).
Cheers