Overview of complex event processing technologies (i)

I have always been puzzled by the understanding of complex event processing (CEP)-Why is this computational model effective and able to analyze and get useful results? Why is it fast? I still haven't found the answer I expected. Unlike the Map-reduce model, Google's paper describes its scene very clearly, or the disruptor framework, which clearly explains why it is fast. In trying to understand the CEP, I am ready to translate some of the articles I think are instructive, and hopefully also useful for everyone. The original article "anoverview of Complex Event processing" is very long, I will divide it into several to translate.

Liu Binhua original translation, reproduced please indicate the source http://www.cnblogs.com/Binhua-Liu/p/5325346.html

What is event handling (processing)?

In the world around us, every minute of every hour of the day, the human brain is bombarded by or in sequence, or at the same time, with endless things, and these events immediately seem to be completely meaningless, even somewhat eccentric, but as more and more things happen, We can begin to understand the relevance and importance of their relationship.

For example, we heard a distant cheer, we saw balloons flying in the air, music began to play, police cars and the appearance of the towed puppet truck, sitting in the car waving people, followed by the ambulance, and today's date is July 4. Individually, these events may mean anything, but together? This is probably a Independence Day Carnival parade!

Our brains can easily determine this fact in the blink of an---although not too simple. As defined by calculation, we can describe a "cruise event pattern" as follows:

One (or more) police cars + rear/rear/adjacent + one (or more) Carnival trucks + one (or more) of the rear/rear/adjoining + a (or multiple) waving person + + (or more) of the rear/adjoining + one (or more) emergency vehicles + where to hear music + today's date is July 4

Your brain is not confined to this mode of work: Send messages, wait until there is a response, and then perform a series of fixed steps to complete a task. As in this example, it is possible to correlate the sequence of events that are taking place with the relevant external factors such as today's date, and to understand that what is happening now is the "parade event pattern".

So, when you learn more about complex event handling (Complex event processing)--The technology we focus on--you'll know how it takes advantage of continuous, flowing, never-ending information from different places to instantly understand what's going on, And what is going to happen in the very near future. This is often referred to as real-time situational awareness (real-time situation awareness). This is associated with business computing

The problem in the computer world today is the proliferation of data. Information from many different systems, to huge quantities, at different times, at different speeds, some of which are important to certain systems, people or processes now , while others can be stored first for later recovery and decision-making. Why do we now face the problem of data proliferation?

Many issues are involved here, but only a few of the main ones are listed here:

• The cost of computing power and complex environmental sensor equipment has become less expensive

• Increased networking capabilities and become smarter

• Many different functional computing modules (financial systems, production systems, sales systems, etc.) are decomposed and rewritten to meet the needs of the more business.

• New computer solutions require the proliferation of partners and customers beyond the enterprise itself, bringing more and more data sources and other inputs into the system.

• Computing architectures such as service-oriented architectures (SOA) are becoming more and more successful, bringing more complex and reusable ecosystems to the system.

• Big Data exploded, and the word is now widely used to describe large-scale, high-speed, diverse, unstructured, information from social networking sites, mobile phones, and many other sources.

• The growing expectations of the IT team for business operations – requiring them to respond more effectively and in real time to market conditions

Our progress and these complex systems have led to a growing "dumping" of information into computer applications, and we have reached a "tipping point" where traditional point-to-point or request-answer solutions become ineffective, difficult to maintain, and non-extensible.

The operation of a company may be affected by events that occur instantaneously, not only in an internal, understandable, "comfortable" small world, but possibly from outside events such as the "Internet of Things"-real-time sensor devices can measure and report many situations, including " A sudden increase in the temperature of a food storage facility, or a sea container tracked by a GPS location, is detected by the sensor to be unauthorized out of the box.

The direct impact on a company's operations may also come from the growing social media changes to the global business environment, such as Twitter, instant messaging applications, and so on. Millions of people at the same time can simultaneously give a bad comment on a new product, emphasizing a product design that needs to be modified urgently. This is bound to affect profits, and even significantly affect the value of enterprises. As a result, companies are inevitably struggling with positive or negative events.

In the past, it may be traced back 15 of years that commercial applications must not stoop to the fixed methods, structures and interfaces (such as relational databases) used in the computing technology that was available at that time, and the information had to be inserted and stored statically before the user could parse and respond again. The common implementation of the traditional JEE (Java Enterprise Edition) application server is to look for an initial request sent by the client application, and then in most cases a large amount of code logic is used to process the request from beginning to end, You can then return the response to the client. These technologies are and will continue to provide solutions that are more batch-based and have poor real-time properties. The new, lower-latency, faster, memory-based middleware products are now on the market.

An event-driven architecture (Event-driven)-based system is inherently more agile and better "armed" to handle these types of situations-dealing with events that need to be interpreted and handled immediately, across the entire business infrastructure and the "silos" of many business units (such as finance, production, and sales). These types of systems, once they detect changes in the external and commercial environment, can contact the context and execute the policy. Instead of pre-defined tasks that are performed every night, even those tasks that need to be manually run each time.

As related to big data, the issue of significant growth over the next few years-event-driven technologies (especially complex event processing) that can be used to capture, manage, and process in an acceptable timeframe, could provide a faster, more "time-to-event" approach to higher-level "intelligence" and decision-making, The original data stream.

Therefore, the advantage of event-driven technology is that in dealing with data proliferation, traditional data processing methods, instead of describing information as real-time events (which can happen anywhere), provide the ability to intelligently decompose events, route events, filter events, and correlate events, so in most cases Decentralized events can evolve into comprehensive, solid, understandable business events that enable businesses to observe, control, and adapt to changing situations in a relatively better way.

Cond...

Overview of complex event processing technologies (i)