IBM's eLiza: Self-healing IT

IBM's Project eLiza aims to supply enterprises with self-managing infrastructure that will free up IT budgets and relieve administrators from tedious network chores. Maggie Biggs explores eLiza's inner workings.

As the server market heats up, a new initiative from IBM--eLiza--aims to supply enterprise IT with self-managing technology infrastructures that promise to reduce TCO dramatically.

The new eLiza technologies let IBM products fix themselves, taking the human element out of the picture. As a result, network administrators can focus their efforts on more critical work. eLiza involves not only all of IBM's products and services, but those of other technology providers such as Nortel Networks and BMC Software.

This isn't the first we're seeing of eLiza. A mid-1960s IBM project--known as ELIZA--focused on communication between humans and computers and was the precursor to much of today's artificial intelligence technology. IBM's new eLiza initiative also focuses on supplying intelligence, but for enterprise infrastructures rather than between humans and computers. For example, IBM is equipping its server platforms with intelligence that can be used to automatically allocate computing resources on the fly as business conditions change.

The four promises of eLiza--to self-configure, self-heal, self-optimize, and self-protect--could increase availability, reliability, and security across the enterprise. IBM expects its self-configuring technologies to enable dynamic resource additions and changes, such as server processing allocation, while self-healing capabilities will detect failing components and take the necessary recovery steps.

eLiza's self-optimizing capabilities enable IT to define acceptable service level agreements (SLAs) for the enterprise, which are based on business goals. The SLAs can then be acted upon by hardware, software, and services that contain the self-optimization technologies.

Today, managing complex security processes at any large company can absorb budgets quickly, given the amount of expertise required and the number of technologies involved. The self-protecting functionality promises to integrate security processes across platforms while also centralizing security management. ELiza technology is designed to detect and report intrusions as they happen, as well as provide backup/recovery and defend systems against unauthorized users. Ultimately, you could rely on eLiza sufficiently to possibly reduce your reliance on vendors and third-party support organizations, letting you actually reduce the cost of hardware, software, and support that you need to purchase.

Given eLiza's broad vision and large implementation scope, you might naturally wonder if eLiza technologies might affect enterprise infrastructure performance. According to Anthony Befi, vice president of eServer Design at IBM, "overall system performance is a very important part of eLiza. The company is designing eLiza functionality so it's implemented in the lowest layers of hardware, operating systems, and network infrastructure. This design approach should increase manageability without impeding performance."

eLiza today
IBM is beginning to deliver portions of its vision through its hardware, software, and service offerings. In January, IBM released an update to its IBM Director systems management software that lets its eServer xSeries line of Intel-based servers fix themselves. And last week at CeBit, IBM announced its mainframe-class Intel server, the eServer x440, which has been in development for three years. Another example involves xSeries servers that are being outfitted with Light Path Diagnostics, which helps administrators more easily troubleshoot server component failures. Lights are associated with specific components on the servers. These lights illuminate at the time of a failure. Components include memory, processors, hard drives, power supplies, and cooling fans. Administrators can quickly and easily identify a failing component via the physical lights--not a screen display--potentially without even running diagnostics. Light Path Diagnostics are similar in concept to the lighted indicators on a copier machine that show the location of a paper jam.

IBM's Unix-based pSeries servers are also being equipped with eLiza capabilities, such as logical partitioning. The latter has been available on iSeries servers for a while. It lets Unix administrators split a system into multiple partitions; each partition has its own independent processors, memory, storage, and OS. For example, you might run the same e-commerce application in two partitions in order to help balance server load. If one partition failed, the second partition could temporarily assume the load while eLiza would perform diagnostics and perhaps apply a remedy. The application would remain available despite the failure.

Another compelling eLiza implementation is the Intelligent Resource Director (IRD). Currently, IRD is available on IBM zSeries mainframe servers, where it automatically allocates processing power on the fly. IRD can manage the resources of multiple OSes executing on the same server as if it were a single resource. The IRD-equipped server can dynamically balance activity and allocate resources based on business objectives. For example, if the OS powering your e-commerce solution suddenly hits a traffic peak, IRD can automatically detect a workload change and re-allocate processing power from other concurrently running OSes (known as virtual servers) on the server that don't require as high a processing priority. You can expect IRD to be included on other IBM server platforms in the future.With eLiza, IBM is also including its software with new features, particularly with respect to Tivoli management applications. For example, Tivoli's SecureWay Risk Manager, will let IT integrate and manage information from a variety of security technologies, including intrusion detection systems and firewalls.

The Tivoli SecureWay Policy Director and Tivoli Identity Director will work together to increase security levels across the enterprise. The former will let you create and implement a security policy for your entire enterprise, while the latter will let you control access to systems, applications, and data in a centralized and automated fashion. These self-protecting capabilities reduce the amount of administration needed to manage security throughout the enterprise. IBM expects to add more eLiza capabilities to Tivoli and its other middleware, including MQSeries and WebSphere, in the near future.

IBM is also adding elements of eLiza to its service offerings. For example, a portion of its e-business Management Services now includes Workload Process Performance, which lets businesses set priorities for resource allocation to more efficiently process transactions. Finally, IBM has added eLiza functions to its Virtual Help Desk service, which now can automatically perform diagnostics to fix software problems in a self-healing manner. IBM will continue to add more eLiza functions to its other services in the near future.

Not new, just broader in scope

Certainly, self-managing technologies are not a new idea. Portions of the Oracle 9i database are self-managing; for example, the application can manage its own database workloads. Likewise, Sun Microsystems' Jini technology can be used to make devices self-managing.

However, IBM's ideas about self-management in the eLiza project are much broader in scope than the implementations of Oracle, Sun Microsystems, and others. By tackling the enterprise technology infrastructure as a whole, IBM hopes to make self-management a standard part of hardware, software, and services across the industry in the future--which could ultimately mean a sizeable TCO reduction for enterprises. Such a reduction could be possible within two to five years, if IBM delivers on eLiza and convinces the rest of the industry to adopt it.

Do you think self-managing servers would benefit your network? Talk back below or e-mail us with your thoughts.