Network Computingwww.networkcomputing.com

Availability measures whether the messaging system components are properly functioning. Because the messaging system is dependent on communications resources, you need to have your fingers on the pulse of the underlying network components to determ ine their health. Congestion on the network, routing bottlenecks, a link failure or a downed server has an impact on the availability and the response time of the messaging system. The more messaging-related components the management product can monitor, the more proactive your support will be.

Reliability is determined by measuring how many messages are successfully delivered between users over the course of the measurement period. The key indicator here is the number of nondelivery notifications typically resulting from misaddressed, timed-out or corrupted messages. All the products we looked at offer totals for messages sent, and most keep track of failed messages. Other factors influencing reliability are the number of directory lookups per second and the number of replication updates per second.

Accessibility is based on how many users are successfully requesting services of the messaging system at the same time. It also looks at the ratio of the total number of successful logons to the number of failed logons. Other useful indicators are peak logons so that you get trend information for busy times and simultaneous connections, required for capacity planning and license compliance.

Delivery time, also referred to as response time, is the average period that messages take to travel from one message store to another and be available to a user. AppManager, TOPPER and MailCheck solve the problem of measuring response time by generating messages that are sent between message stores. The time that the message takes to be delivered or bounced is tracked and reported.

Note that response time now extends beyond the enterprise, and it is important to consider external user expectations for mail delivery. Organizations are more often using collaborative computing processes, such as workflow, to reach out and bring the customer into the business process (for example, extending the loan application process out to the customer).

The capacity, size and volume of various messaging components has a direct be aring on the overall performance of the system. Systems choked by too many users or constrained by inadequate hardware will grind performance to a halt. Taking a look at the total number of message accounts, the number of simultaneous connections, the size of the message store, message and log file size and server indicators-- such as CPU, disk and memory utilization--over time will help you spot trends and justify upgraded or additional hardware and software licenses.

Determining Affordability Although not strictly performance-related, the affordability of the messaging system also should be tracked. How much does it cost to send messages through your system? How much does it cost to provide a particular level of service to users? Is the system performing within the budget outlined in the SLA?

You will want to have several options available for costing message traffic, including subscription-based charging and charges allocated based on the usage of a messaging resource. Managers need to have in -depth charge-back information for the entire enterprise as well as for various departments and messaging resources.

The notion of a premium service rate for designated users is becoming more prevalent as products offer better accounting, billing and charge-back capabilities. Multiple levels of service often are needed between the same end points, even if the products don't deliver it. From the user's perspective, why not send all your mail out as urgent if it costs the same? If there are no constraints against sending a 10-MB attachment during the busiest hour of the day, why not do it? With a premium rate structure, heavy users would pay more for sending high priority, urgent mail, for using a greater amount of message storage and for sending large files during busy times.

If the management product provides the means to assign and gather both metrics and costs, it is much easier to perform a cost-benefit analysis to justify improvements to the messaging system. You'll want the management system to gen erate two types of reports, one for the system administrators showing troubleshooting type of results (downtime and bottlenecks, for example) and one for management showing performance and costs.

Reporting the Results How can you present your findings on the performance and cost of the messaging system to management and users? It is crucial that the workload, metrics and costs generated by the messaging management products be documented and reported. These reports determine SLA compliance by documenting both the objective for the metric and system capacity as well as the actual performance. Also, reporting enables management by exception, which reflects changes in the messaging system, such as those brought about by a merger, or novel feature like embedded hotlinks to a particularly popular Web site.

All the products we looked at offered a bevy of well-formatted and practical reports, including graphical, text, online and printed reports as well as those that could be e-mailed or published on t he Web. Graphical reports can be dynamic or historical showing real-time status or timely consolidations of enterprise servers--it's your messaging system. With judicious selection and use of performance measurement tools, you'll be better-equipped to fight brush fires, become more proactive in managing the enterprise system, model new features and functionality, and determine the overall cost of providing a messaging service for your organization.

Nancy Cox can be reached at ncox@nwc.com.