Comments on: Complex Event Processing: a technology evaluation check-list

By: Paul Vincent

Paul Vincent — Thu, 26 Aug 2010 10:22:38 +0000

Thanks Keith - but we don’t really want the TIBCO CEP blog comment system to be used to advertise other vendors’ events!!! Nontheless, thanks for the invite and good luck.
Cheers

By: Keith Dunlop

Keith Dunlop — Tue, 24 Aug 2010 21:56:29 +0000

Invitation to register for Sybase CEP Webinar on September 14th, 2010

…

By: Paul Vincent

Paul Vincent — Mon, 08 Mar 2010 23:01:36 +0000

Hi Tina: Thanks for the feedback:

- On “latency” and other performance metrics, of course everyone “has good performance” :). Then again, I didn’t specify this was a “paper evaluation checklist” so although I should do a “software trial checklist” later I will add performance to the platform component of this list.

- I would map “application management” to “operations and management”; “impact analysis” / “auditability” map somewhat to verification and validation (see the earlier comment response to Hans and James…). On the other hand, it is up to end user organizations to assess their need here, so I will add that in under business interfaces.

Cheers

By: Tina Groves

Tina Groves — Mon, 08 Mar 2010 19:06:17 +0000

The criteria seems fairly complete though latency appears missing (implied in the scalability criteria?). I agree with Hans that capabilities related to application management such as impact analysis, auditability and so on warrant inclusion. The scoring, as you say, is relative to the use case and, I would add, target environment. “CEP Opportunity Analysis and Assessment” by IDC’s Fleming and Silverstein (Dec 2008) does a nice job describing how priorities shift according to the use case.

Thanks for sharing.

By: Paul Vincent

Paul Vincent — Sun, 07 Mar 2010 22:52:50 +0000

Hi Hans, James: I would claim that “impact analysis” in Event Processing could be viewed as 2 different activities:
- monitoring and prediction (what will happen downstream of incoming events)
and
- “what if” analysis of such monitoring and prediction (normally covered by “simulation” in CEP).

The first (monitoring and prediction per se) is usually a pretty common CEP role. We are trying to predict what will happen based on event patterns occuring. Examples would be predicting electricity grid transformer failures, or manufacturing process quality test requirements (both TIBCO BusinessEvents use cases). I think the industry is at an early stage of providing “business tools” to handle the complexity of the development and specification of the temporal pattern logic required for CEP…

The second, simulation, is “relatively easy” in simple use cases (like the test facilities in TIBCO BusinessEvents Decision Manager, and in many other BRMS tools), but likewise more difficult to achieve effectively for event sequences involving time and current state, especially for a business user. Indeed, outside of simple “event simulation” components in CEP tools, I’m not aware of any business suite that does this effectively today. Probably you would need to automate support for some technique like Combinatorial Testing.

On the other hand, we do see occasions where test cases are generated by rules (and separate rules than those providing the CEP!) to try and provide good coverage - indeed this is used in some of the TIBCO BusinessEvents provided examples. Overall, test-based development and the use of techniques like event generators for testing I would consider as “best practices”!

Cheers

By: Paul Vincent

Paul Vincent — Sun, 07 Mar 2010 22:29:26 +0000

Hi Opher - thanks for the advice - yes, I will simplify this to just GVs (Global Variables) “etc” as there are many ways of doing a global context…
Cheers

By: Hans

Hans — Sat, 06 Mar 2010 14:38:50 +0000

Would you not say that impact analysis is highly dependent on the use case? Or is there a general kind of impact analysis that’s not coming to my mind right now?

By: James Taylor

James Taylor — Sat, 06 Mar 2010 01:03:24 +0000

Paul
A nice list but I think you are missing a key element in the business interface - impact analysis. Business users, in my experience, don’t want to test or verify they want to see the impact of a potential change in business terms.
JT

By: Opher Etzion

Opher Etzion — Fri, 05 Mar 2010 15:23:11 +0000

Hi Paul. Good work in collecting all criteria, my main issue with scorecards that they are based on the assumption that there is a universal scoring, as some of this criteria may or may not apply to certain application, also it is sometimes the case that an event processing system need hybrid of capabilities, and it is difficult to define exact scoring. I believe that the way forward will be to apply a collection of building block that can easily work together so each customer can pick what it needs, and also extend without problem, as applications evolve.

cheers,

Opher