11.13 Management frameworks

 < Day Day Up > 



After your organization begins to grow, administrators will spend more and more time grappling with servers and event logs. The manual tools are good enough to monitor Exchange, but the amount of manual intervention required quickly becomes a problem. Without automation, you run the risk that you:

  • Miss early warning signs for problems. For example, you miss a transient –1018 event, which may point to a storage problem, or you miss events flagged when virtual memory fragmentation occurs.

  • Experience avoidable outages caused by disk space exhaustion on important volumes.

  • Fail to notice signs of security breeches, such as users logging on to mailboxes without permission.

  • Fail to notice gradual performance degradation of a server until it reaches a point that affects users.

  • Fail to notice gradual or heavy accumulation of messages on email queues or network problems that cause mail flow to halt.

  • Fail to collate reports from multiple servers that may point to common problems across the infrastructure.

  • Fail to notice that a server is down or that a critical service (e.g., the Information Store) or component (an individual Store) is offline.

Management frameworks help system administrators by automating system monitoring to identify potential problems and then suggest solutions. Unless you are very paranoid and spend all your time monitoring servers or are just lucky enough to be looking when a problem happens, management frameworks are always faster than people at detecting and reporting problems. Generally, you have to install agent software that runs on the servers you want to monitor and management packs that interpret the data. The agents collect information from various sources and collate it in a database or other repository, which you can then query with management consoles.

Management packs are specific to each application and provide the frameworks with added intelligence over simple analysis by applying rules that search the collated data for warning signs and then compare problem symptoms against their knowledge base to arrive at a recommendation for the administrator to action. Management packs typically include rules to process the data, thresholds that help determine when problems exist, offthe-shelf reports, and scripts. For example, the Exchange 2003 management pack for MOM (which Microsoft provides as part of Exchange 2003) includes rules that look at over 1,700 different events that the various Exchange components (Store, Routing Engine, MTA, System Attendant, protocol stacks, etc.) log. The events are tagged with their severity level and pointers to possible solutions in the MOM knowledge base. Various notifications are supported, including messages to pagers and so on.

There are three major management frameworks suitable for Exchange 2000/2003:

  • Microsoft Management Operations Manager (MOM)

  • HP OpenView

  • NetIQ AppManager

Many factors influence the choice of which framework is best for your organization, including:

  • If you have already deployed a framework to manage other applications such as SQL/Server, it makes sense to extend that framework to cover Exchange. This is especially true if your company has invested in any customization, such as the development of knowledge scripts used by the monitoring agents.

  • If you operate a heterogeneous computing environment spanning multiple operating systems, it is best to select a single framework that can span all platforms. For example, HP OpenView is especially powerful when you need to monitor UNIX and Windows systems and network components.

  • One of MOM's major strengths is its use of Microsoft's own knowledge base and information gleaned from developers, and you might want to benefit from that experience. It is also good to be able to add to the knowledge base to reflect your own experience.

  • You may be able to use the prepackaged tools to assist in storage, server, and network capacity planning and replace other utilities that you deploy for this purpose.

  • Cost of deployment and maintenance. You have to purchase the central monitoring software plus the other components that you need, including agents and management packs tailored for the applications that you want to monitor, and then license them for your organization. You also need to pay for ongoing support, updates to the knowledge base, and so on. You probably need different versions of the software for each version of Exchange you operate.

In all cases, you should conduct a test to ensure that the chosen framework performs reliably and meets your needs. Of course, you can write your own management and monitoring tools using the Exchange WMI providers, other APIs such as ADSI, and a mass of code that gathers and interprets system data. Large Exchange organizations have existed since the earliest days of the product, so the need for automated monitoring of large numbers of servers is not new. It is reasonably common to find monitoring code, but few companies are willing to take on the development and maintenance when they realize that off-the-shelf packages are available. For this reason, while Exchange continues to provide better programmatic access to its data with each release, it is less common to attempt to write custom code to attack the monitoring problem. A curious symmetry indeed!



 < Day Day Up > 



Microsoft Exchange Server 2003
Microsoft Exchange Server 2003 Administrators Pocket Consultant
ISBN: 0735619786
EAN: 2147483647
Year: 2003
Pages: 188

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net