A framework-based network management system (NMS) doesn't necessarily have to consist of scores of modules that support a supercomplex, hypereclectic computing and network environment. For lots of networks (or if you want to get your feet wet), the framework plus a few modules may be all you need. We call these tools framework express, or framework lite.

Getting to the core of network management by choosing only those modules that support your key devices, servers and applications can be an effective, affordable, productive and smart approach to using a framework-based NMS.

The ideal framework-express package includes a central management and monitoring piece, to which you add a few modules that recognise and manage specific devices, servers and applications. Each module blends seamlessly into the overall NMS, has a small footprint and is easy to use. Modules work together to manage everything, automate administrator tasks, process SNMP alerts (traps), discover the network and diagnose outages and performance problems. The perfect package offers useful reports, scales well, is pervasively platform-neutral and enforces good security.

To find a system that meets our criteria, we invited vendors to submit systems to our lab for testing. We tested HP's OpenView Network Node Manager 7.5, OpenView Operations 7.5 and OpenView Internet Services 6.0; BMC Software's Performance Manager Console 7.5.20, Distribution Server 7.1.21 and Performance Manager Portal 1.2.00; and PerformanceIT's ProIT IT Operations Management Software 4.0.

Computer Associates, which had just acquired Aprisma at the time of our tests, said it needed to think about the positioning of Unicenter vs Aprisma's products and declined our invitation. IBM's Tivoli division, after initially agreeing to participate, backed out of the tests and said a suitable product wouldn't be ready until June.

We awarded HP the Clear Choice Award for OpenView's excellent network discovery, root-cause problem analysis, task automation, responsive and intuitive user interface, and scalability.

How we did it
Our test environment consisted of six routed Fast Ethernet subnet domains and a T-1 Internet connection. We ran each product's server components on four-way Compaq ProLiant ML570 900MHz computers with Pentium III CPUs, 2GB of RAM and six 18GB SCSI RAID drives. The operating system was Windows 2000 Advanced Server with SP 4. Each subnet's 25 client computers were a mix of Windows NT 4.0, Win 2000, Win 2003, Win 98, ME, XP, Red Hat Linux 7.0 and Macintosh platforms. The relational databases on the network were Oracle 8i, Sybase Adaptive Server 11.5 and Microsoft SQL Server 2000. Win 2000 and NetWare 5.1 shared files, while Internet Information Server, Netscape and Apache software served up Web pages. An Agilent Advisor protocol analyser decoded and displayed network traffic.

We evaluated each product's ability to manage, administer, update, monitor, report on, diagnose, troubleshoot, reset, reconfigure, audit (inventory) and secure network devices, server computers and client computers. Virtually all our testing took place across WAN links.

In our tests, we administered users, groups, servers, clients, routers, switches, remote storage and DSU/CSUs. We tested the sending of SNMP alerts, as well as the processing of incoming alerts. We produced reports to show device and computer status information, inventory results, network usage trends, security breaches, availability and uptime information, network baseline information and graphical maps of the network. We tested any special features the product offered, and we also looked for scalability, security, ease of use and task automation.

Barry Nance runs Network Testing Labs and is the author of Introduction to Networking, 4th edition and Client/Server LAN Programming.