Hardware-based load-balancing mechanisms can also be easily
integrated as the endpoint of a service, and can represent a fa?§ade for a server cluster with
0 B skup, Heyer, & Marx G??mez
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
a hardware load balancer handling the incoming requests. Naturally these techniques can
also be combined.
To be able to effectively handle faults, several approaches can be implemented; the important
concern is whether service faults on the actual service level (e.g., complete service failures
due to unavailable servers, etc.) or on the logical level (e.g., service errors, etc.) are examined.
The communication layer can easily enhance the fault tolerance of a Hyperservice-based
system on the service level: A simple protocol handler extension can be integrated in order to
allow the resubmission of service requests if a service endpoint fails to respond. This behavior
can be enhanced by integrating a regular availability check into the service/resource registry
to be able to filter out or deactivate temporarily-unavailable endpoints and thus reduce the
overall network load.
Pages:
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501