Highly Availability for Multi-Tier Architectures

With the increasing pervasiveness of eServices in all areas of life resulting in a high dependency on them. This results in a need for highly available services. High availability is attained by introducing redundancy in the underlining Service Oriented Infrastructures (SOI) typically by means of replication (either data or process replication). We look for contributions at how to introduce this redundancy in multi-tier systems to attain high service availability. We look for solutions for both single and multiple tiers including federation of replicated tiers.
One of the most important issues that should be dealt with in the approaches for high availability is data consistency in the advent of failures and concurrent access by multiple clients. High data consistency typically results in poor performance. Approaches dealing with relaxed level of consistency are therefore also of interest in which the consistency/performance tradeoff is identified. Another important issue that is called for is how to overcome real life problems when implementing replication such as how to enforce determinism, how to deal with sources of non-determinism, etc.
We also invite for contributions addressing performability. That is, how to deliver the same performance during failures and recoveries as in the failure-free execution. Non-intrusive solutions for fault-tolerance and recovery, such as online recovery, are therefore sought.

The scope is high availability for SOI. High availability solutions for any kind of service (and underlying server) are welcomed, including typical tiers of multi-tier architectures such as web servers, application servers, and databases. The high availability of specific servers such as composition (e.g. BPEL) engines, service directories, etc. are also in scope. High availability for SOI deployed in any kind of network is in scope: LAN, WAN, high bandwidth, mobile, ad hoc, ...