Glossary

Complex Event Processing

Processing and analyzing event data in real-time to distill high level information and insight from low level events. The word complex refers to the fact that simple events are filtered and combined to form complex events.

Throughput

Typically measured in messages per second, throughput measures the capacity of the system to process messages in real-time without getting overloaded and without significant queuing that would introduce latency

Latency

Latency is the delay or lag time introduced by a system. In measuring the latency of a CEP engine, latency refers to the time from when an incoming event message arrives until the processing of the event is complete and the result is delivered to a client application.

Scalable

The Aleri CEP engine can scale to handle high message rates, large data sets and complex data models. It is a multi-threaded 64 bit architecture that takes full advantage of available resources (CPUs, cores, and memory) to scale to the needs of the application. It can also be deployed across multiple blades or individual machines, using either clustered or cascaded configurations. This provides for virtually unlimited scalability.

Enterprise-class

Applications that are deployed as part of the enterprise infrastructure performing mission critical tasks have different needs than application that run on a desktop or those that may run on a server but that perform a non-critical task. Enterprise-class applications demand a robust, scalable architecture designed for continuous operation and that has features such as high availability and security.

High Availability

The Aleri server can be deployed in a high availability configuration that consists of two live servers running as a redundant pair. One server runs as the primary, the other runs as the secondary. In the case of failure of the primary, the secondary will automatically take over to ensure no loss of service.

Adapters

Adapters are frequently used to integrate the Aleri server with data sources and data consumers. Adapters use the Aleri pub/sub API to connect to the Aleri server, and use the interface of the data source or data consumer to retrieve/receive data or deliver data respectively. An example would be the Aleri JMS adapter which subscribes to messages delivered via a JMS message bus and streams them into the Aleri server. A range of pre-built adapters are available from Aleri and custom adapters can be built by Aleri, by Aleri customers, or by third parties.

Relational Operators

Standard relational operators are things such as filter, join, and aggregate. These are the operations that are easily performed using an SQL Select statement.

Multi-dimensional Analysis

Multi-dimension analysis of data refers to viewing data that is aggregated or summarized across various dimensions. Let’s say your data consists of individual orders for the last year. You might want to see how many orders were received in each month – time is a dimension. You might want to see how many orders were received for each product – product name (or ID) is a dimension. Dimensions can also have hierarchies: time for example can be grouped by week, month, quarter, year. Product might be individual product, product family, product category. The hierarchical nature of the data also allows for something often referred to as “drill down” – where you start with a high level summary and then move down the hierarchy of a particular dimension to view more detail.

OLAP

On-Line Analytical Processing, or OLAP, refers to a class of applications that allows a user to interactively perform multi-dimensional analysis against a set of data. With an OLAP tool, the user can interactively explore the data, summarizing the data across different dimensions and drilling down to see a greater level of detail. OLAP tools are designed around a data architecture that can respond to the user’s queries quickly, without making the user wait, as opposed to running a complex query against a relational database where it might take minutes or even hours to deliver the results.

Dataflow Programming Model

Dataflow programming models a program as a directed graph of data flowing between operations. Dataflow programming is, among others things, a technique for making parallel processing easier. It is very effective in an event driven architecture where data enters and flows through the directed graph, with each node in the graph performing an independent operation on the data

Directed Graph

A series of nodes (or, in this context, operators) that are connected by directional arcs that direct the flow of data between the operators

Data Persistence

In the context of event processing, the term data persistence can be used to refer to several different things. A CEP engine will typically retain data in memory for use in processing future events. Aleri provides the option of disk-based persistence for retained data, which ensures that data is not lost in the event of a system failure or other shut-down scenario. Any retained data that has been configured for persistence will be restored when the system is restarted. This is also referred to as full state recovery since the system is fully restored to the state it was in before the shut-down or failure. Data persistence is also sometimes used to refer to the collection of data into an historical database, though this is not how Aleri uses the term.