Tmax Engine

This chapter introduces the Tmax engine, a system component of OpenFrame/Base, and describes how it integrates with OpenFrame.

1. Overview

Mainframes were initially designed to handle large batch processes that may run up to several hours. They are not suitable for handling a large number of small transactions that can be called thousands of times and processed within few seconds or minutes.

In the late 1960s, IBM introduced IMS and CICS to handle such transactions, and these products are generally called TP monitors. Technically, a TP monitor is not part of an operating system. It is a type of a batch application that manages transactions.

Many of these early TP monitors were associated with a specific database system where one server machine is dedicated to processing the transactions. With the growth of online services over time, recent TP monitors use multiple programs across distributed C/S systems.

2. Tmax Introduction

Tmax, TmaxSoft TP monitor, guarantees transactions across heterogeneous computing systems, balances loads, and handles transactional errors. The following are some of its features.

  • Compliance with the X/Open DTP international standards for distributed transaction processing

  • Automatic management of the API and system interface in compliance with the OSI-DTP international standards

  • A transparent view of the services available in a distributed system composed of heterogeneous machines

  • OLTP support

  • An optimal development environment by maintaining ACID (atomicity, consistency, isolation, and durability) properties during transaction processing

3. OpenFrame and Tmax

OpenFrame is based on the Tmax engine. It inherits many Tmax features, such as multi-node clustering, load balancing, and failover, guaranteeing the same level of reusability and reliability as mainframe.

The following are OpenFrame features inherited from Tmax.

  • Multi-node clustering

    A Tmax domain consists of one or more nodes (machines) that share an environment and are linked together in a peer-to-peer relationship. With access to any of the nodes, you can use all services provided within the domain.

    Multi-node clustering allows applications (previously run on a single mainframe) to run on multiple, distributed Unix machines in a stable OpenFrame environment.

  • Load balancing

    When service demands from Tmax clients spike, the following three load balancing methods can guarantee optimal system performance and resource utilization.

    • System load management (SLM)

    • Data-dependent routing (DDR)

    • Dynamic load management or balancing (DLM)

    Tmax load balancing features fully extend the use of each Unix machine in a clustered OpenFrame system.

  • Failover

    Tmax’s failover function guarantees transaction continuity even when an unexpected failure occurs.

    The following are Tmax’s failover methods.

    • Failover through service backup for hardware (node or network) failure

    • Failover through process restart for software failures (process down)