The end-of-year analyst report (aka “earnings call”) is where financial analysts listen to and question the corporate officers of a company like TIBCO on their end-of-year report card. TIBCO’s 2009 report (recorded Dec 22 2009) delivered by the CEO, COO and CFO had some comments relevant to the CEP business:

• TIBCO innovation in event store-and-forward approaches: the release of TIBCO ActiveSpaces
• “[I]t is actually substantially more valuable to have just a little bit of the right information, at the right place, at the right time and in the right context than having all the information in the world six months after the fact.” This followed 3 customer examples doing TIBCO-based CEP (a “Western states utility”, a “major Asian bank”, and a “major Indian mobile company” -  covering energy, finance and telco industries).
• On EDA: “[T]here’s going to be a systematic shift from transactional to event driven architectures … transactions don’t pickup threats and opportunities … there is a systematic shift, it’s like a change that’s taking place …”
• On BPM: “We’ve seen a number of situations where a customer actually went away from say an ERP CRM type implementation as in the case of the Asian bank and went through an event driven inbound marketing approach.”
• On BI: “The traditional BI players are largely reporting systems … they allow you to analyze and mine data after the fact and what we do is we look at streaming events before and allow you to anticipate what’s going to happen as it is about to happen. Then, we also with our Spotfire product have taken the visualization of it through a whole different level. I think those are the two elements, the predictive real time nature versus the reporting after the fact nature and the visual technology that goes with it.”

Computing (a UK IT weekly) has an article called Tools of the Architect’s Trade. In it, Gregor Baus from Air France describes some aspects of their EDA system for “smart boarding” - utilizing TIBCO CEP technology, as it happens. Note that “smart boarding” is about intelligent passenger operations, and is nothing to do with CIA interrogation techniques…

And with that, TIBCO’s newest offering, Silver is off and running. 

Announced yesterday at the NOWonline show, it seems to be getting a good bit of attention in the press, analyst and blogosphere communities.  eBizQ picked up on the announcement and commented on its use of CEP in the automation of cloud-app-balancing. As for me, my head is a bit cloudy at the moment, from all the fuss.

So what is Silver, and what does it have to do with CEP? 

Everything. 

TIBCO Silver is new software infrastructure for “cloud” computing.  A “Silver” lining for the clouds you might say. 

And why is this important for CEP? 

Because it’s an infrastructure product that embeds a CEP engine in order to solve problems related to governance (managed access, security, privacy and adherence to regulations), and scalability (uses SLAs to automatically scale up / or down as needed).  The kicker is that it’s automatic, so both the governance and the scaling is accomplished inherently through embedded monitoring, management and event-decision-action rules rather than manual intervention and programming -which AFAIK, is an achilles heel for current cloud products being introduced. 

This should be an interesting announcement for developers of different types of Business2Consumer or Consumer2Consumer apps that are likely to vary widely in resource requirements. The embedded governance allows for various levels of authorization, authentication and encryption policies to be dynamically configured. This is important because some services should be open to everyone and some services, well, just shouldn’t.

As in most cloud architectures, and not counting those who simply put the cloud moniker in front of their latest software product, there is no software to install or hardware to procure or provision, which reduces the barrier to develop and deploy rapid IT solutions (whether that’s infrastructure, platform or applications)

TIBCO Silver is currently in Beta. It will be interesting to see the deployments when they start rolling out.

Ashwin used this great slide at our TIBCO BusinessEvents QLGroup best-practices session this week, courtesy of James Gosling’s blog and credited to Peter Deutsch.

Essentially everyone, when they first build a distributed application, makes the following eight assumptions. All prove to be false in the long run and all cause big trouble and painful learning experiences.
1. The network is reliable
2. Latency is zero
3. Bandwidth is infinite
4. The network is secure
5. Topology doesn’t change
6. There is one administrator
7. Transport cost is zero
8. The network is homogeneous

Arnon Rotem-Gal-Oz did a follow-up white paper that goes into more details.

Apart from classic Complex Event Processing (event abstraction across time, source, type and attribute) applications for situation-awareness type problems, TIBCO BusinessEvents is also finding uses in sophisticated / dynamic event-driven processes that cannot be fitted easily / entirely within representations such as BPMN, BPEL, etc. Of course, there are many other ways of deploying complex processes across services: consider the wsper project, for example, that aims to exploit and combine WSDL, SCA and BPEL [*1]. Their provided example, interestingly, also maps very nicely onto an EDA using a combination of state model + rules, which have the added advantage of declarative rules that can handle any out-of-process (i.e. complex exception) event [*2] - something BPEL can’t handle [*3].

Notes:

[1] I don’t agree with the wsper metamodel (documented here), which defines an “event” as a subclass of operation, and “state” as an attribute only of event. This seems far too restrictive, although it might fit wsper’s SOA perspective.

[2] Consider the (unfortunate) case whereby, in the supplied HR example, your job candidate “expires” (RIP) in the middle of this process.  You would not want your business process to attempt communication with the dead - you need some flexible, declarative, cross-process exception mechanism. This is typically handled by a “rule”…

[3] On the topic of BPEL we also note one effort looking at retrofitting “event processing” constructs into BPEL. Why not event-driven COBOL, too, one wonders? A more interesting project would surely be standardize the event modeling level (e.g. extend BPMN for CEP, or map CEP to BPDM), rather than trying to force event processing into BPEL (insert here visions of pigs in party frocks). No doubt this topic will come up for some (lively) discussion at the EPTS meeting on CEP standards next month…