Moving away from Mainframe to Commodity – How?
Mainframe (Z/OS) based systems running COBOL programs are legacy systems in many organizations. These are planned to be replaced with low cost commodity servers running Java or .Net based systems, saving the cost of the expensive mainframe MIPS and COBOL-based development.
Using GigaSpaces XAP can simplify the migration effort from mainframe based systems and reduce the cost of the legacy applications. In addition, having GigaSpaces XAP act as a front-end layer for mainframe based systems may boost the system performance and improve the overall system response time on peak load.
GigaSpaces' ability to deploy, manage and scale services along with the data (that can be partitioned and replicated across multiple commodity machines) will enable your mainframe applications to access GigaSpaces XAP's In-Memory Data Grid (IMDG) with minimal re-factoring of existing application code without having to introduce additional third party products, dramatically reducing implementation times and minimizing incremental costs in software licenses and hardware.
GigaSpaces Intelligent Mainframe Front-end Architecture
GigaSpaces XAP provides an extremely flexible persistence layer (known as the mirror service) that enables transparent communication between the GigaSpaces IMDG and virtually any type of back-end application or database system.
When used with a database, the Mirror service is one of the primary reasons allowing GigaSpaces XAP to overcome database locking issues experienced on peak load periods. The Mirror service offloads the database access, since the IMDG operates as the primary interface to the application data while handling persistence as a back-end durable ordered activity, delegating in-memory transactions to the database running on the mainframe.
Any access to the data done primarily from the IMDG using one of the standard interfaces GigaSpaces XAP supports (POJO/Spring, JPA, JDBC, key/value, or Document APIs). If the desired data item cannot be found within the IMDG, it will be accessed through the database running on the mainframe, retrieving the relevant data item, loading it into the IMDG to be reused for subsequent transactions and passing it back to the client application. This approach saves the need for accessing the mainframe on every application data access by using an in-memory layer that may scale on demand.
Controlled, Reliable, and Optimized Mainframe Access
XAP's Mirror service has a central coordinator for all back-end store updates, enabling you to batch data and persist in-memory transactions via a continuous background access to the mainframe where the frequency of access is pre-configured. This allows the system to minimize the number of mainframe connections and interactions reducing MIPS consumption while controlling the data consistency level and synchronization between the in-memory representation of the data and the copy on the mainframe.
Many mainframe-based applications that perform nightly batch jobs drive a large number of data updates being made to back-end stores. In this context, GigaSpaces' inherent ability to maintain transactional integrity is critical. In-Memory transactions can be fully committed preserved in multiple different physical locations using GigaSpaces' high-availability mechanism, and ultimately persisted to the database with zero risk of the mainframe and GigaSpaces being out of sync for a long duration.
For more details see the Mainframe Integration Best Practice.