Zaptron Systems

Technology:
High Availability Computing


High Availability Computing & NGN Protocols

A White Paper
Zaptron Systems, Inc.
 

Introduction

For many years computers have been controlling systems that provide critical services where continuous availability and data integrity are essential. The scale, the availability requirements, and the level of data integrity vary widely across the spectrum of such services. Consequently, the techniques and implementations that provide the high availability (HA) are also equally varied. However, the history and evolution of these applications has added extra dimensions of complexity and variance to the HA implementations. Often the development has been ad hoc and narrowly focused, and has usually led to proprietary solutions that are fragile and expensive to support. Much of the same technology has been re-invented time and time again, but in ways that are incompatible and impossible to reuse. 

With the widespread adoption of open systems and more horizontally-focused component suppliers, there is growing expectation for an open system approach to provide products and services that address the needs of high availability in a variety of application spaces. This is one of the motivations of the foundation of Service Availability Forum (SAForm).


Service availability vs. system availability

l        System availability means that if a resource (e.g., a card or an entire server) should fail, another is available to take its place.

l        Transaction data is often lost during switchover to the new component. Service is interrupted.

l        Service availability not only must achieve the highest levels of system availability (99.999% or above), but must preserve transactions and uninterrupted connections despite any resource failures

     

Customer Requirements for Open HA Systems

    Standardize the design of commonly used HA components and techniques

    Encourage the development of Commercial-Off-The-Shelf (COTS) HA components

    Allow competitive environment to improve product features, cost, and performance

    Allow application developers to focus on core value-add development

    Reduce the HA application development time and cost

    Improve the portability of HA applications between hardware platforms, operating systems and middleware

     

     

     

SCTP (Stream Control Transmission Protocol

SCTP is a new IETF transport protocol for reliable message-oriented data transfer.

It can be used anywhere TCP would be used.

Comparing to TCP, SCTP provides high availability, better data delivery efficiency, stronger security.

 

How is SCTP different than TCP?

Directly support for multi-homing, providing link (or path) -level redundancy for high availability.

Multi-stream capable, a way to escape "head-of-line” blocking.

Message boundaries are preserved

SACK is built into the protocol.

Heartbeat/Keep-alive mechanisms are integral part of the protocol.

•Un-ordered delivery as an option.

•Message time-to-live option.

•Security Cookie mechanism to protect against "SYN" attack.

•Path MTU discovery built-in.

•Better extensibility

 

Where is SCTP being used ?

 

    Multi-homing of SCTP protocol.

 

RSERPOOL Introduces a Session Layer into IP Architecture

RSerPool Features

Copyright © 1997-2005, Zaptron Systems, Inc.