Introduction

In a networked environment high availability of services is very important. Creating redundant content is a diverse topic and each application has its own nuances. This discussion will be made from the point of view of an Internet Services Provider where a typical application is to create a fault tolerant HTTP Server. This has different requirements to providing files for a group of workstations though much of the discussion will be applicable to this and other applications too.

There are two key issues involved in creating redundant content, replication and access.

1.

Replication involves duplicating data by some means so that a failure on the box where files are housed will not mean inaccessibility.

2.

Access involves providing a client with requested files whether this be of a local disk or across a network.

This discussion will cover freely available software solutions under Linux. Commercially available solutions such as the AFS and DFS[Tra] file system are interesting however I would much rather concentrate options that anyone can run free of charge. I will also not be touching on hardware solutions such as hardware bus sharing [HM98] or RAID disk arrays ¹ as the software aspects of this issue are already numerous.