Lakeview Adds Autonomic Features to HA Product Line
May 31, 2005 Dan Burger
When you realize the labor costs involved with setting up, running, and maintaining computer hardware and software surpasses the cost of the equipment itself by factors that sometimes reach as high as 20, is it any wonder why autonomic features create so much interest? Lakeview Technology, an IBM Premier Partner and a major ISV in the iSeries market, has recently loaded its two high availability products–ha1 and ha Lite–with autonomic features designed to simplify the process and reduce the long-term costs.
High availability, or managed availability as Lakeview likes to refer to it, is the consistent and predictable access to any data or applications whenever, wherever, and however users require them. In a world where planned and unplanned downtime can cost a company an arm and a leg–and, worse yet, customers and business deals–high availability and 24×7 operations are taking on increasing importance. And where HA was once only affordable to the corporate giants of the world, it is now being implemented by organizations that clearly fit into the small and medium size business (SMB) category. (IBM has helped this cause by offering dedicated high availability servers at a price point that encourages high availability projects. For more information on this, see “IBM Delivers Model 810 iSeries for HA Server”.)
The differences in the IT requirements at the large enterprise compared to the smaller company may be vast and the difference in IT department staffing may be just as dramatic, but what both organizations have in common is a need for simplicity. When it takes additional staff to monitor new software and hardware, the return on investment stretches further into the future and often beyond the realm of possibility. So Lakeview’s approach to HA now emphasizes autonomics. It comes from years of experience with large enterprise customers and the recognition that auditing and monitoring HA software has personnel tied up with tasks that prevent them from applying their expertise in more productive ways.
Businesses that are using or considering HA solutions need to get their end users over to another system–a backup server–that allows them to be productive. This occurs primarily during planned downtime. So high availability becomes a means of getting everything needed (data and applications) to the backup server.
“HA is often talked about in the context of data replication,” says Glenn Van Benschoten, one of Lakeview’s technical gurus who work in the company’s Rochester, Minnesota, research and development facilities. “Data replication is an important component of HA, but data is only a piece of the HA solution. There are ‘environmentals,’ which are wrappered around an application and its data on the system. For example, there are communications and networking factors that determine how users get access to the production or the backup systems.”
How these communications are managed plays a big role in high availability.
Security is a major component of the environment. Data and objects may physically exist on the target server, but if the authorization list, the security information, the user profiles don’t precisely match the production machine, it will cause a delay in the switch over. These delays require additional time and human resources, and, therefore, they can be costly.
“There needs to be checks for any type of error conditions,” Van Benschoten says. Examples of potential errors include the loss of key attributes that keep that environment running, network links that may have gone down, assurance that security information is correct on the target machine, and whether all objects and files are in sync.
This is the checking, monitoring, and auditing process that MIMIX ha1 and MIMIX ha Lite have added as autonomic features.
Van Benschoten provided examples of items that are autonomically audited. “Applications use database files, data queues, data areas, IFS files, system objects, configuration objects, and user profiles, which all need to be controlled, managed, and monitored,” he says. With the new enhancements to MIMIX, “all the replication of these things is done under the covers.” Not only audited, but autonomically corrected when errors occur.
MIMIX relies on remote journaling as the transport mechanism for the five data types it was designed to handle: database files, data areas, data queues, IFS screen files, and large objects (LOBS) that are attached to the database files. Other objects can’t flow through the remote journal, so a separate autonomically audited transport mechanism is used to apply them to the target server.
MIMIX products have included auditing capabilities, but in the past the end user was responsible for taking action to fix any problems the audit identified. That method didn’t assure problems were taken care of quickly. There was always the possibility the user did not see the audit alert in a timely manner or maybe that person had other things to complete before taking care of the problem. Either way it had the potential to compromise availability.
“That doesn’t fit what the end user wanted,” Van Benschoten says. “They want something that takes care of itself.”
Lakeview wrote the code for autonomic procedures based on what it determined to be the best practices in a given situation. That recommended response became the default policy for the autonomics. Van Benschoten describes it as “wrappered, policy-driven autonomic behavior around what is detected.”
“These are routine, mundane things that required manual response before,” according to Van Benschoten. With the autonomic approach, action is taken to correct problem and that action is stored in a log for later analysis. All the autonomic (default) settings are customizable to some degree, or a user can override the settings if necessary, so it is not a strictly “one-size-fits-all” technique. “Our approach is that 80 percent of the customers would probably handle the fix this way and we make that the default,” he says.
The goal for Lakeview is to provide a self-running, self-managing, hands-off, lights-out solution. It sounds like pie in the sky, but the more tasks that are removed from human responsibility, the greater the efficiency and cost effectiveness. Still, the HA solution cannot be expected to be 100 percent set-and-forget.
“I wouldn’t try to tell you that it can screen for the total set of possible problems that could come up in the universe,” Van Benschoten says. “Where we are today is not going to capture every one of those. We went after 80 percent of the stuff that is causing the most aggravation. We audit and correct those. But we will get better as we add even more autonomic behavior into the product. We are not making a lights-out guarantee for every customer. Eighty percent of customers cannot expect to buy HA and set it up and forget it. I don’t think the state-of-the-art HA is anywhere close to that today. But for a smaller set of customers, they can truly set it and forget it (in terms of monitoring). If there is something that needs attention the product will notify them through e-mail or a pager or cell phone. But that person who responds does need to know something about the HA environment.”
Pricing on MIMIX ha1 and MIMIX ha Lite is tier-based. A license is needed on any machine that sends data to a remote target. Most customers buy a pair of licenses. Current MIMIX customers will gain the autonomic features without an additional charge. The company did not provide specific pricing, however, both Lakeview products are available now with the autonomic features.