Application Requirements Drive Container Design

Abstract

There are many possible implementations of the JAIN SLEE standard. Different implementations provide different availability guarantees, have different cluster architectures, provide different ACID semantics, and exhibit different latency and throughput performance. Essentially it is the characteristics of the application and requirements for the application that dictate the necessary features or capabilities in a JAIN SLEE product. It should be noted by the reader that there are many different types of applications that are run by todays Network Operators, Enterprises and ISPs. These applications can differ significantly in the requirements for their processing. For example messaging applications (i.e. cell phone based text messaging) and Call Control applications have very different latency, throughput and availability requirements.

In this section we use the typical characteristics of 'Call Control Applications' and 'Enterprise Applications' as an example to outline the difference characteristics that different containers provide. Call Control Applications are applications that are involved in running phone calls within a signalling network. Enterprise applications are typical EAI or database applications.

Typical Call Control and Enterprise Application Characteristics

	Communications	Enterprise
Invocations	Typically asynchronous - Events such as protocol triggers - Events occurrences mapped to method invocations	Typically synchronous - Database, EAI systems - RPC Calls
Event Granularity	Fine-grained events High Frequency	Course-grained events Low Frequency
Components	Light-weight fine-grained objects Short transient lifetimes - Rapid creation, deletion	Heavy weight data access objects Long persistent lifetimes
Data Sources	Multiple data sources - Location, context information - Provisioned data, cached from master copy	Database servers - Definitive master copy Back-end systems
Transactions	Light-weight transactions - For state replication demarcation - Faster completion and more frequent	Database transactions - Slower completion and less frequent
Computation	Compute-intensive - Processing is resource invocations and events	Database access intensive