The K2 server can be configured to
run across a cluster of computers thus providing higher
performance and a higher degree of reliability. The clustering
is accomplished by using a federated Trading Service,
Fault Tolerance and Management System. The federation
of these modules is organised in a non-hierarchical way
i.e. all the federated parts are equally constituted.
This architecture ensures that there is no single point
of failure and no overhead in maintaining the hierarchy.
A client accesses the cluster
through the federated Trading Service. The federated Trading
Service runs in as many instances as there are nodes, but
every one of those instances contain the complete offer information
of all the nodes of the cluster. The client can go to any
of these Traders to request a service. The Trader will respond
with an offer that is chosen out of the complete cluster.
This architecture not only improves the performance in terms
of how many service requests can be served in a certain amount
of time but also the number of clients that can be served
truly simultaneously as there are multiple initial entry points
into the system.
The communication that is
used to build up the cluster is based on a Reliable IP Multicast
Protocol that works within a LAN segment. The IP multicast
makes the inter-federation communication in the order of magnitude
more efficient than a replicated point to point communication.
Moreover this technology ensures that there are no scaling
up limitations for the cluster because of the inter-federation
communication. The costs for the communication ranging from
2 to 256 nodes within a LAN segment are equally expensive.