Previous | Table of Contents | Next |
No matter how good the features of any product or set of products, if the system is not reliable, people cannot depend on it. If the system is perceived as unreliable, people will use alternative forms of communication.
To satisfy user needs, the IT department should separate internal customers from external customers. Internal customers are those that help provide a service. They may be IT management, support personnel, or networking staff they could be considered an internal supplier.
Because of the nature of most organizations, internal customers are both customer and supplier. They need to be provided with the means to supply a service. For example, IT management may need to create step-by-step procedures for the operations staff to carry them out. If the information technology group cannot satisfy the requirements of internal customers, it probably will not be able to satisfy the needs of external customers.
External customers are the end users. If they are in sales, for example, external customers may include the enterprises customers from other companies. It is the job of the IT staff to provide external customers with messaging features, functionality, and reliability so they can do their job.
It is helpful for network managers to know how other enterprises have implemented messaging systems. The next few sections describe the various components of the infrastructure, common deployment architectures, and how to plan future deployments.
Often messaging systems are sold with the emphasis on what the end user sees. Experienced network managers know that this is the least of their problems. The behind-the-scenes components, which make the individual systems in an organization work as a near-seamless whole, include:
Network Services. The network services required for a messaging rollout involve connectivity between:
It is not unusual to have one network protocol between a desktop device and its server and a second protocol within the backbone server/gateway/router environment. Servers may communicate via WAN protocols such as TCP/IP, OSI, DECnet, or SNA, and the desktops may communicate over a LAN protocol such as IPX or NetBIOS. WAN connections may occur over continuous connections or over asynchronous dialup methods.
The network administrators greatest concern is loss of network connectivity. It is important to understand how it happens, why it happens, how it is discovered, and what needs to be done on an application level once connectivity is restored.
If the network goes down, E-mail will be faulted. Weekly incident reports should be issued that cite direct incidents (i.e., an E-mail component failure) and indirect incidents (i.e., a network failure) as well as remote site issues(i.e., a remote site lost power). Such information can help to clarify the real problem.
Message Transfer Services. The message transfer service (also termed the message transport system) is the most visible part of the messaging infrastructure. The message transfer service is responsible for moving a message from point A to point B. This service consists of one or more message transport agents and may be extended to include gateways and routers. The most popular services are X.400 and SMTP international standards, and IBMs SNA Distributed Services (SNADS) and Novells Message Handling Service (MHS) proprietary industry standards.
X.400. More widely used in Europe than in North America, X.400 is popular because it:
SMTP. Simple Mail Transfer Protocols allure is its simplicity. Addressing is easier and access to the Internet is relatively simple compared with establishing an X.400 connection. Because it is simple, there is not much that can go wrong. However, when something does go wrong, it is usually monumental.
Directory Services. The directory service is critical to a companys E-mail systems, but it is also problematic. The problems are a result of the difficulty in keeping directories up-to-date, resolving redundant or obsolete auto-registered entries, and failures of directory synchronization.
The directory serves both users and applications. End users choose potential recipients from a directory. The directory should list enough information for a user to distinguish between the George Smith in accounting and the George Smith in engineering. Some companies include in their directory individuals who are customers and suppliers. The ability to distinguish between internal users and external users is even more important in these cases.
Management and Administration Services. Management refers to scheduled maintenance and automated housekeeping procedures that involve system-related tasks such as reconfiguration and file maintenance. The constant I/O on messaging components leads to disk and sometimes memory fragmentation. Regular defragmentation procedures, including repro/reorg, tidy procedures, and checkstat and reclaim, are required. Whatever the environment, such procedures should be done more often than is recommended to prevent problems from occurring.
Alerts and Alarms. Alerts and alarms are extremely helpful because the system can tell the user if there is a potential problem. Alerts generally refer to warnings such as too many messages in queue awaiting delivery. Alarms are a sign of a more serious problem, such as a disk full condition.
Mail Monitoring. Mail monitoring is typically an administrative function. One way of monitoring a system is to send a probe addressed to an invalid user on a target system. On many systems, the target system will reject the message with a no such addressee non-delivery message. When the initiating system receives this message, it indicates that mail flow is active.
Timing the round-trip provides a window to overall system performance. A message that does not return in a pre-established timeframe is considered overdue and is cause for further investigation.
Reporting. Reporting is used for capacity planning, measuring throughput and performance, chargeback, and statistical gathering. At initial implementation, network administrators will generally want to report breadth of coverage to demonstrate the reach of the infrastructure. Breadth can be measured by counting users and the number of messaging systems within each messaging environment.
Exhibit 1. Implementation Scenarios | ||
Enterprise | ||
---|---|---|
Single | Multiple | |
Single Department | Single System | Similar Systems |
Multiple Departments | Dissimilar Systems | Cross-Enterprise Systems |
Previous | Table of Contents | Next |