Intro

Introduction

Why do we build Cluster Servers?

Todays information systems plays a central role in business processes. To guarantee trouble-free functioning of these business processes an organization depends on the continous availability of the complex and heterogeneous infrastructure as well on the manifold of interconnected and interdependent network services.

To prevent information systems becoming unavailable because of a Single Point of Failure, Cluster Servers are built and deployed. These servers ensure that the service remain available to client applications in the event of any faults. The cluster represents a virtual Server consisting of two or more independent servers called cluster nodes serving one or more services which can be used by clients over the network.

The virtual cluster server gets addressed and accessed over a virtual IP address which is assigned to one of the cluster nodes. On failure of the active server of the cluster the Virtual IP-Address gets reassigned to one of the still available cluster nodes which in turn will provide the service and answers to request from clients.

The cluster nodes of the cluster automatically detect the failure of a cluster node or a service. The service gets reassigned to another cluster node transparently and gets started on that new cluster node. This process is called Fail-Over.

For what does SRRD stand?

SRRD stands for Service Routing Redundancy Daemon. It's a Unix daemon implementing a cluster server.

Service Routing Redundancy is a new and different way to build cluster servers which implement redundant services.

What can SRRD do?

SRRD implements a fully featured Cluster Server based soley on the OSPF routing protocol.

SRRD is using a new OSPF feature named OSPF Opaque Link State Announcements as proposed in RFC 2370. These Opaque Link State Announcements allow to flood arbitrary data through an OSPF routing domain.

SRRD is fully configurable over the web and supports SSL and PKI client and server authentication. It implements cluster server features like Service Groups, Service Dependencies as well as Critical Services.

The participating cluster nodes may reside on different networks (LANs) accross the whole OSPF routing domain allowing fast failover between distant cluster nodes.

Check the Features page for all available features. On the Screen Shots page you can find screen shots of a 5 node cluster. If you wonder what signals a service primitive can receive, check the introduction to Service Primitives page.