Much has been written about how IBM's autonomic technologies have made the iSeries easier to manage. When it comes to high availability solutions, however, self-managing, self-healing, self-auditing, and adaptively self-learning solutions were considered out of reach. Not so any longer.
Recent and not-so-recent innovations in HA software research and development make it clear that, when designed properly, autonomics features and functions based on best practices can save significant management time and maximize IT resource efficiencies. Autonomic features can also enable automated, smart detection and correction of HA issues (replication and synchronization issues) as well as other issues that arise throughout the HA environment. The IT organization benefits from a healthy, efficient HA solution as well as the ongoing confidence that the solution is always ready to switch to a backup server when wanted or needed -- even as the IT environment changes over time.
What's behind HA autonomics and how do they make HA for the iSeries even better? If you have invested in an HA solution sometime ago and want to improve its performance or are currently evaluating HA solutions for the iSeries, you should know how autonomics now make HA solutions self-managing, self-healing, self-auditing, and adaptively self-learning. And they require much less time to manage.
- Autonomics should be policy-driven and auditable. Autonomic functions should be both predictable and configurable. Ideally autonomic functionality should embody the best of "HA Best Practices," based on the cumulative experience of IBM developers and field-tested software solutions and the support people behind them. The goal would be to create HA software that would be its own "expert" and automatically take pre-approved actions that can later be reviewed and tweaked if necessary.
- Autonomics should find and repair issues immediately. The most effective autonomic capabilities "stay on duty" to significantly minimize operator intervention. For example, some HA software will audit replication processes after the fact. That's too late. Autonomic features should be able to audit the replication process "inline" in real-time during replication.
Today's autonomic features now enable policy-driven self-healing as well, so any issues that arise during replication are corrected immediately -- even as users stay active. Of course, you will always want the capability to go back and perform an audit of the replication processes on demand, but to truly keep production and backup servers in sync, inline audits and self-healing will solve issues before they cascade into larger issues later.
- Autonomics must be adaptive and proactive. By incorporating HA best practices, today's autonomic functionality can now offer adaptive recovery. That means the HA solution will automatically select the most efficient and effective way to fix errors based on the type of problem identified. In other words, autonomics lets an HA solution become very selective in its approach to any issue that arises.
Today's autonomic-driven HA solutions also use built-in, proactive exploratory diagnostics to expand the investigation of an issue and self-heal beyond just the identified problem. In other words, today's autonomic features can go beyond a replication problem and provide some insight into the possible cause, which may or may not be related to the HA processes.
- Autonomics must be efficient. How efficiently the HA solution performs its autonomic tasks will affect your total cost of ownership (TCO). For example, today's autonomics features should be able to identify and fix a specific problem rather than needing to work with an entire file to fix a specific data or object issue. Rather than the sledgehammer approach of yesterday, autonomics now enables a more precise, laser-beam-like fix. That ensures more efficient use of server resources and lowers TCO.
- Autonomics need to identify and solve recurring issues, too. Most HA solutions will keep finding and resolving the same issue, in the same way, again and again and will not alert you to this fact -- wasting time and resources. Autonomic features can now identify recurring issues before they can escalate to larger problems. Autonomic features now make it possible for the HA solution to alert the operator through autonomic smart notification features -- pager, e-mail or instant messaging -- about a problem the solution has automatically fixed more than four times in 24 hours.
- Autonomics must provide confidence that your HA environment is error-free. Look for an HA solution that will automatically initiate error correction, as well as detect, apply and manage changes within your entire environment, including triggers, nulls, constraints, LOBS, commitment control, and table issues. Some HA solutions provide some or a few of these, but today's autonomics features make it possible for an HA solution to provide automated, vigilant observation and correction of elements within the IT infrastructure. In addition, when looking at the level of autonomics available, you will want to understand what is automatic and what must be done or corrected manually.
- Autonomics should manage remote journaling. While IBM's Remote Journaling technology has been a boon to HA software vendors, unfortunately it doesn't manage itself. This job can be more time-consuming than it at first appears. Today's autonomic features can eliminate this chore and enable you to get the full advantages of Remote Journaling without the time involvement.
Look for autonomic features that will create Remote Journaling journals and the links between them for you, automate the day-to-day tasks and switch management of Remote Journaling, reconfigure and manage replication processes, change and delete journal receivers, and even restart failed Remote Journaling links -- the communications resources used in replication. Should something happen, today's HA solutions should be able to automatically enable failover to alternate communications paths. Autonomic features provide these capabilities.
Data replication is just one piece of a HA software solution. Today's most efficient and effective HA solutions must be able to manage replication effectively, but also be able to manage the surrounding IT processes, such as communications and networking factors, to ensure user access to the data and applications on the production or the backup systems.
Today's most advanced HA solutions will be able to do this for you by incorporating autonomics features. While the features perform complex tasks behind the scenes, the whole point is to lower the amount of time it takes to manage the HA environment without reducing the confidence you have in the solution. Autonomics-enabled HA solutions hide all of the complexity of HA to lower the TCO and make management as easy as possible. In fact, it should be able to lower your time involvement to just a few minutes a day.
About the author: Peter Sa is vice president of MIMIX Services at Lakeview Technology Inc.