Database Reference
In-Depth Information
Introduction to the complex event
processing engine
There are two terms that are generally used in conjunction; they are
Complex Event Pro-
cessing
(
CEP
) and
Event Stream Processing
(
ESP
).
Well, in theory, these are part of a technical paradigm that allow us to build applications
with dramatic, real-time analytics over streaming data. They let us process incoming events
at a very fast rate and execute SQL-like queries on top of the stream of events to generate
real-time histograms. We can assume that CEP is an inversion of traditional databases. In
the case of traditional DBMS and RDBMS, we have data stored, and then we run SQL
queries over them to arrive at results, while in the case of CEPs, we have the queries pre-
defined or stored and we run the data through them. We can envision this with an example;
let's say I am running a department store and I would like to know the highest selling item
in the last one hour. So if you look here, the query we are about to execute is pretty fixed in
nature but the input data isn't constant—it changes at each sale transaction. Similarly, let's
say I run a stock holding company and would like to know the top 10 performers over the
last 2 minutes every 5 seconds.
The preceding figure depicts the stock ticker use case where we have a sliding window of 2
minutes and the stock ticker is sliding every 5 seconds. We have many practical use cases
for this nowadays, such as:
• Fraud detection patterns against
Point Of Sales
(
POS
) transactions
• Top
N
in any segment
• The application of deep learning patterns to stream data from any source
Now, having understood CEP and its need at a high level, let's touch upon its high-level
components:
Search WWH ::
Custom Search