If you can set up an IP address, this product will manage your network. It may not be the best thing since sliced bread, but InfraTools Network Discovery, or IND, is a top-notch network-management appliance, and it toasted the competitors' buns. This product hides much of the down-and-dirty implementation and maintenance involved in network management, but that's not what makes it a winner. Accuracy has to come first, followed by just the right amount of useful information. You need to know what your network is doing; what it has been doing; and, if the product has the crystal balls, you may get a prediction of things to come.
IND is a fourth-generation Linux-based appliance first offered by Loran Technologies and recently purchased by Peregrine. Peregrine, known for asset-management software, sees IND as part of its larger strategy to inventory everything on the network. IND fits into this strategy because it inventories nodes to the port level.
Ever since vendors attempted to document and map port-level connectivity, useless port-level information has been produced by network-management products. Why? Because SNMP bridge MIBs, Cisco Discovery Protocol and even proprietary MIBs are not consistent in their implementation of module/port-to-MAC (Media Access Control) address mappings. IND has been around long enough to have workarounds to address most of the SNMP funk hanging over our networks.
Although Network Node Manager, OpenRiver and Spectrum attempt to read bridge, repeater and proprietary MIBs to get at the morsels of MAC to module/port data, they do it with varying levels of success. IND's edge here is its ability to fix changes fast. How? Most of these offerings rely on an API or complex object definitions that are rolled out when the product is released. Peregrine, however, has a proprietary scripting language for poring through MIBs to assemble needed pieces of information. The company tracks new inconsistencies constantly, then updates and revises the scripts that put the MIB pieces together. Because IND is an appliance, there's no waiting for updates, recompiling, rebooting and applying patches. Updates are provided as soon as they become available, automatically, as part of the purchase price.
'Exceptional' Management
Exception management traditionally has been the domain of trap- and event-processing. Correlation of traps is dangerous work -- only the skilled, the few, the marvelous need apply. IND completely sidesteps this problem, however; it's the only product we tested that does no trap-/event-processing.
So how can the product be any good at exception management? IND came in a plain brown wrapper -- a nondescript rack-mount Intel box with a keypad. We punched in an IP address, mask and gateway, then we went away for a couple of days. The south of France is nice this time of year.
After a few days, we fired up a browser -- the only console IND has -- and got a Java-applet-rendered map and Health Panel. If you're worried about a Java-only console's functionality, don't be. This one is pretty good, with right-click option menus, progress indicators and multiple paths to access functions. The Health Panel displays alarms and warnings on three categorized events: line faults, device faults and metrics (changes and repair outages).
Line faults are notification events, such as line breaks, utilization, delay, collisions, broadcast, errors and packet loss. Devices that are unreachable are identified by IND as device breaks, contributing to the device-fault category. Metrics include information on devices that have been moved as well as Peregrine's proprietary Network Early Warning System, or NEWS. NEWS indicates collected utilization and packet-loss statistics and the number of days until thresholds are violated. Also included in metrics are statistics on the MTTR (mean time to repair) and MTBF (mean time between failures).
Although the product offers a fully functional events browser that displays filtered and historical views of events, we were satisfied to stick with the Health Panel, a tight, intuitive application. The Health Panel let us change the thresholds for alarms and warnings, as long as we were logged on with administrative rights. Contrast this with Spectrum, in which thresholds are part of the device properties, and with NetworkIT, where separate applications spawn multiple windows.
A unique part of the Health Panel is the Time Warp function. Not only did IND's automatic data collection and data archiving let us look back at the network -- it let us look forward. Presented are time periods, such as yesterday, last week, last month, six months or up to a year ago. When we selected one of these periods, the minimum, maximum and average MIB statistics -- and all the other Health Panel data -- were displayed.
Reporting on the past is one thing; predicting the future is a bit more difficult. IND doesn't claim to prognosticate with 100 percent accuracy. Instead, it delivers a confidence rating along with its predictions, which look between one month and one year ahead. The NEWS predictions are affected by these forecasts, so we could estimate when interfaces would run out of gas, bandwidthwise.
IND's basic report package includes network inventory, line utilization, exception summary and detail data. Business Reports, an add-on pack, offers executive summary, WAN, LAN, device and support reports. Reports are displayed in HTML and presented in real time. Although the reports can be printed and include links to devices reported on, for the most part users can't modify the scheduling and formatting. The reports are well-documented and intuitive, but we did wonder why the LAN backbone report showed a significant decrease of availability when we knew no such problem existed. As it turns out, the report includes ports from routers and switches, not what we consider the backbone.
Not All Smooth Sailing
One notable disadvantage is that, unlike NetView, NetworkIT and Spectrum, IND does not let administrators create customized device groups. We would have liked to create a backbone group, but we'll have to wait until IND adds user-defined reports, a feature the company said is coming in the next release.
We ran into a pair of additional problems. In the first case, our NetScreen Technologies firewall prevented the MIB browser from working off network. This was because the NetScreen firewall prompted for a CRLF (carriage return line feed) instead of passing the traffic. The second anomaly was a random resetting of many network devices to the name "Spirit." At first we thought it might be a nonmalicious tag by a student. In both cases, IND found the problems and in the "Spirit" case offered up a fix.
InfraTools Network Discovery 4.0, $50,000 for 1,000 nodes, $190,000 for 10,000 nodes. Available: Now. Peregrine Systems, (800) 638-5231, (858) 481-5000; fax (858) 481-1751. www.peregrine.com
REPORTS
ANALYTICS
Follow 
