GRID AND AUTONOMIC COMPUTING Grid computing and autonomic computing can be considered complimentary technologies. Both are evolving, being developed, and implemented at the same time. The technologies are utilizing many of the same components, such as Web services, applications, networks, open standards, and storage. The configuration, deployment, maintenance, and management of existing networks is and has always been a complex endeavor. Growth and rapid implementation of new business applications have fueled even more complexity, which is reaching the point where these systems become unstable and almost unmanageable. The skill requirements for managing these complex environments are extensive and can include detailed knowledge of the network topology, distributed data behavior, load balancing, performance tuning, and system optimization. As we identified earlier in this book, highly skilled IT systems administrators are hard to recruit and even harder to keep. They are expensive, but very necessary, resources. Many have already reached the limits of what they can manage today. With system complexity increasing, the need to introduce self-managing systems such as autonomic computing is becoming a paramount priority. This analysis can be taken a step further by introducing the prediction that grid computing will only gain widespread acceptance in the global marketplace by integrating autonomic functionality. Grids need self-managed, self-configuring systems within the networks to manage their devices and resources with little or no human intervention. Many of the following features will be included: -
System configuration management— The OGSA CRM is used to model the system and its resources. Generated events are handled according to user set policy. -
Job execution management— This covers time-, priority-, and space-based scheduling of jobs. In case of application failure, jobs are retried based on applicable policy and priority. -
Resource management— Dynamic and flexible resource management is essential. At the same time, resource isolation between different jobs is crucial, not only for access control but also to ensure that there are no unexpected performance dependencies. -
Infrastructure management— Backup and recovery, failsafe mechanisms, restart procedures, storage allocation and maintenance, and report printing and distribution are all necessary to infrastructure management. -
Clustering features— To handle resource failures and provisioning, features for adaptive resource allocation should be provided to enable autonomic management. The actual behavior should be based on client-specified policies. -
Infrastructure services— Some of these services are end-user management, accounting and billing management, pay as you go/pay what you use, system logging, and tracing. -
Security management— This provides a single sign-on authentication service, with support for local control over access rights and mapping from global to local user identities and security profile handling. -
Grid monitoring— This is an integrated information service distributed across the grid to monitor resources that provide information about the state of the grid infrastructure. The services should be based on the Lightweight Directory Access Protocol (LDAP). -
Network management— This provides an interface to TCP, UDP, and file data I/O. It should support synchronous and asynchronous interfaces, multithreading, and integrated GSI security. -
Provide services— A service that implements a variety of automatic and programmer-managed data movement and data access strategies, enabling programs running at remote locations to read and write local data. -
Workload management— This can be described in three tiers: -
Integration: The need to manage HTTP, IIOP, JMS, and other requests as they move through the enterprise. -
Load balancing: Managing the workload based on the resources available at any given time. -
Failure identification: Describing and documenting what went wrong, and mentioning all of the resources involved. |
