Network administrators like few things better than tools that make network monitoring and administration easier and more efficient. After all, keeping the network running keeps the customers happy, which keeps the boss content, no matter if you are monitoring a small Windows NT network with 400 nodes or a worldwide, 18,000-node heterogeneous environment. Mission-critical means just that, and Hewlett-Packard's Version 6.0 of Network Node Manager (NNM) meets the challenge of supporting mission-critical networks.
Whatever your network consists of, NNM will tell you which devices are up and which are down on the network. HP uses a straightforward combination of tried and true network tools, such as PING, to gather node information and monitor the status of network devices, lay out a network topology, and keep track of network configuration changes.
NNM is robust, but it faces competition from products such as Cabletron's Spectrum and IBM's Tivoli NetView. NNM may not be the prettiest face in the crowd, but I like it because it is direct. Also, if your network is heterogeneous, you will appreciate that NNM looks and acts the same whether it is deployed on Unix and Windows NT.
If you are using NNM 5.0, NNM 6.0 introduces a variety of tools that make it an attractive upgrade, particularly the Event Correlation System (ECS) within the Alarm browser.
The ECS can monitor and reduce event storms in the Alarm browser by defining relationships between events. ECS analyses events based on previous, current, and ongoing events, and by analysing these, it can replace a large group of events with a single and more relevant event alarm. This was especially helpful when I was initially faced with hundreds of event alarms per day while I was setting up NNM.
I was also pleased to see that a Web interface was added to NNM. Providing a secure front end for distributing views of the network to team members, the Web interface let me customise different views of the network, and gave me some tools that are not available in the NNM console, including a tabular list of network devices, an ECS configuration manager, and the capability to enable or disable scheduled maintenance (to suppress alarms on devices that will be brought down for testing or upgrades). And although the NNM Alarm browser cannot sort alarms, the Web interface Alarm browser adds that capability to the tool.
The front end of NNM is straightforward, consisting of the topology map, which is a graphical layout of the network based on information from the network discovery, and the Alarm browser, which is a listing of events that have occurred on the network.
The topology map lays out the network in a hierarchical fashion and lets you drill down from the routers to switched segments and finally to the interfaces on the nodes themselves. Network status and performance conditions are presented as different-colored device icons, giving instant visual feedback. The Alarm browser provides a concise listing of network conditions and their severity.
Speeding up network discovery
When I started using NNM, the product took an exceptionally long time to rediscover network changes every time I started my machine after the initial network discovery. After installing several new DLLs provided by HP, rediscovery time dropped from more than 10 minutes to less than 1 minute.
Although NNM can discover network devices with or without a seed address, I had NNM initially discover my 450-node network with a seed address. This sweep of my network went quickly because NNM used the Address Resolution Protocol Cache on the seed device that I provided to PING devices. NNM used SNMP information to find manageable devices on my network whether they were nodes, hubs, switches, or routers. NNM also discovered Desktop Management Interface and HTTP devices. Lastly, NNM gathered IPX network information by using IPX PING sweeps.
When discovering, NNM splits network devices into two categories: managed and unmanaged. NNM actively polls managed devices for information. Unmanaged devices exist within the NNM database and are displayed in the topology map, but are not polled. NNM can page or e-mail you if a managed device, such as a server, reaches above a certain CPU or other specified threshold, or if it detects a problem with a critical network device.
NNM's strength lies in its family ties -- it is one piece of a complete and extensive family of network, server, and application monitoring tools. A variety of HP products, including ManageX server and application monitoring tool for NT (see our Enterprise Networking Product Review, www.infoworld.com/printlinks), can feed information to NNM.
If you are considering upgrading to or deploying NNM 6.0, be forewarned that it isn't the easiest tool to configure when it comes to modifying filters and polling options, but the initial hassle is worth the end result.
(Victor Garza (71333.153@compuserve. com) is a freelance reviewer in Silicon Valley who worked at the InfoWorld Test Center for more than eight years.)The bottom line: excellentNetwork Node Manager (NNM) 6.0With its new Web interface, Event Correlation System, and data-warehouse component, NNM 6.0 can support any mission-critical network, whether it is small and running only Windows NT, or large with a mix of NT and Unix machines.
Pros: Scales up to heterogeneous, multiprotocol networks; works well with other HP products.
Cons: Both NT and Unix versions require configuration file modifications to customise advanced features.
Hewlett-Packard, (800) 752-0900; www.openview.hpcom.
Price: $US4,995 for as many as 250 nodes; $16,995 for unlimited nodes.
Platforms: Windows NT, HP-UX, Solaris.