If you've tried to choke this down in the past, you may have found you couldn't stomach the amount of money and time required to get payback. Management frameworks offered a delicious combo plate that removed the need to deal with multiple vendors for each management challenge. But the fast-paced changes in networks have rendered these frameworks obsolete. Since last year at this time, dozens of vendors have brought new management products to market. Most are performance-management related with the majority of those designed to manage your e-commerce Web site.
Web Performance
Performance management is low-hanging fruit for vendors and enterprises. Whereas management frameworks have provided the enterprise lots of detail, they've never answered the simple question, "How are we doing?" With performance-management products, vendors can leverage the SNMP and IP standards to provide a high-level view of the network and its applications.
Web performance-monitoring products proliferated this past year. Packages like Hewlett-Packard Co.'s WebTransaction Observer, Holistix Remote Monitor and Response Networks' Response Center represent performance management's second wave, adding Web performance to the mix. More established products, such as Concord Communications' eHealth, Lucent Technologies' VitalSuite, NetScout Systems' nGenius and Trinagy's Trend, are still flourishing. These packages also gather MIB2, RMON, proprietary-agent and third-party data feeds.
An important but often neglected part of performance management is transaction performance in a networked environment. We are not talking about real-time measurements, such as those provided by Web performance-monitoring products. Rather, transaction-performance products, such as Compuware Corp.'s Application Expert and Opnet Technologies' IT Decision Guru with Application Characterization Environment (ACE), work offline and break down all the performance facets of a transaction. They tell you how many turns the client and server must perform to complete a transaction, what latency is built into the network, and how increasing server processing or network bandwidth would speed up the transaction.
Sim City
As network complexity has grown, simulation has become a must. This past year, simulating performance has gotten easier -- a good thing, since until now it has been difficult and expensive. NetRule by Analytical Engines is one standout, offering statistical simulation that is fast, accurate and easy.
Products that simulate transactions on networks don't gather and break down real transactions; instead, they simulate the load that such a transaction would put on a network and a system. If you're thinking one would be useful to the other, you're right. Opnet's ACE breaks down transactions to be fed into Decision Guru, its discrete simulation tool.
Centralized data-center operations require a central point of control. Framework vendors have reworked their applications to deliver more out-of-the-box functions to speed up ROI (return on investment). The old-guard frameworks, such as Aprisma Management Technologies' Spectrum, Computer Associates International's Unicenter, HP OpenView Network Node Manager and Tivoli Systems' NetView, which looked as though they were on the ropes last year, have gotten new life by doing more with less implementation. For example, Spectrum and Network Node Manager have raised their functionality while accelerating their ROI. In addition to Spectrum's Layer 2 root-cause analysis -- that is, correlation across applications, service and infrastructure -- the package includes third-party management modules for equipment from Cisco Systems, Nortel Networks and 3Com Corp. Network Node Manager has added Web reporting and management while maintaining attractive pricing.
At the other end of the spectrum are smaller version-management tools that don't try to be everything to everyone. Rather, they focus on a particular problem, and they do it well. Fluke Networks' Net Tool Connectivity Tester, which makes wiring-closet diagnostics easy, is such a device. Ipswitch's WhatsUp Gold delivers a lot of bang for the buck; for under a grand, it keeps track of what's up and what's down.
Peregrine Systems' InfraTool Network Discovery has gone one step farther in simple deployment: InfraTool delivers a network console via an appliance (read our online review). InfraTool's network inventory is the most accurate among these products. Network-management offerings are going to need this level of accuracy to address the difficult problem of root-cause analysis.
By now, protocol analyzers might seem old hat. In reality, they are critical to network management. These devices still offer the best way to get inside a transaction as it is happening on the network. Using them is the only way to look at every stage of a multitier e-commerce application system while it is running. And there is no way of getting around the need to install an analyzer in a wiring closet to track occasional user problems. Not only do distributed protocol analyzers pick up where standalone analyzers left off, but the devices mean network engineers don't have to get off their butts. Network Associates' Distributed Sniffer System, cousin of the famous Sniffer, remains the best.
Still have an appetite for this stuff? Try some desktop management. The task is decidedly unsexy, as products have been around for years attempting to make it manageable. But Novell's ZENworks for Desktops raises the bar a notch. Integration into NDS, support for laptops, workstation imaging and full support for Microsoft Windows 2000 have made it the best tool to keep administrators out of the black hole of desktop support.
Although desktop management is a mature category, the products have never completely solved the enterprise-asset-management problem. In the next year, a number of vendors will introduce asset-management services to resolve this shortcoming. Peregrine, for example, is in the process of starting up such a service, as is Tally Systems.
Management will always be a challenge. Standards- and Web-based applications have simplified basic monitoring, but to solve the next, much more difficult problem of root-cause analysis, new management solutions are needed. Vendors must adopt and champion standards like CIM (Common Information Model) and SNMP configurations, which are designed to make the management name space and data common.
Send your comments on this article to Bruce Boardman at bboardman@nwc.com.
Network & Systems Management
REPORTS
ANALYTICS
Follow 
