IT professionals want to be able to identify and avoid network performance degradation before it reaches critical mass and blows their business out of the water. The good news is that Eye of the Storm offers a single interface for monitoring critical issues vital to the network operating center (NOC), such as real-time fault and performance issues and historical inventory and network capacity trends. The bad news is that problems with the discovery process can eat up valuable time. In other words, EotS is not perfect. But until a perfect network-monitoring product comes along, it's definitely worth your consideration.
I unleashed a beta version of the new EotS at our Real-World Labs® at Syracuse University and asked it to automatically discover the more than 10,000 devices on the campus network. Although devices can also be entered manually, I decided to give the software a chance to do its magic.
I originally began my testing using a PIII-600 with 512 Mb of RAM, but the powerful EotS soon brought this box to its knees, leaving both the hardware and me screaming for mercy. In order to support a network of our size, I had to bump the hardware up to a quad Xeon-700 server with 4 Gb of RAM. EotS spent the first few hours scanning the campus' class-b subnet, and it ran a week before its databases were fully populated and the software was ready to produce a full set of reports. During discovery, EotS not only mapped the infrastructure devices on the network, but also noted such information as the typical loads on various links, availability of user-defined network applications, unused ports and attached nodes.
The device discovery function of EotS missed occasionally, but it collected and properly identified the majority of devices on the network. Unfortunately, it missed a high percentage of the non-Cisco equipment on the network, including several newer HP switches.
Another snag occurred when, although the databases were fully populated and the software seemed to be fully operational, I received a whopping 470 hits while searching for the EotS server's NIC MAC address. In addition to duplicate MAC addresses, there were addresses from completely different subnets. I contacted support and after sending them an inventory dump file, support sent me a recompiled executable, which lessened the problems. But even with the new executable, a few devices had duplicate MAC addresses. Entuity is aware of this issue, and by the time this article is published, it should have a fix for this problem.
The user interface (UI) is fairly intuitive. I wouldn't say it allows you to staff your NOC with a slew of network operation newbies, but it certainly does a lot to empower lesser-experienced individuals. The software's interface is served in a combination of Web pages and Java applications, but fear not -- it was stable and feature rich. Overall, the performance of the UI was very good as well. I was able to do most of my testing via the Internet using a cable modem connection without much of an impact on the speed or reliability of the interface.
In addition to a satisfying layout and intuitive navigation, the EotS UI is customizable based upon user login and customizable domain filters. The domain filters allow you to sort out objects based on the device attributes and can be used to help create different views in the component viewer. By creating customized filters and views, I was able to delegate viewing rights based on user login, which means it is possible for service providers to separate customer data. Likewise, corporate users are able to delegate responsibility by location. I could input my own customized MIB variables easily, allowing me to collect data on those unique or lesser-known variables.
Hit or Miss
The device ports I tracked down in order to verify the accuracy of the data collected showed EotS listed most of the devices and their interconnections correctly. However, there were still a number of unrecognized devices, such as the HP switches. The Proliferate utility attempts to resolve this problem by walking the unrecognized device in an attempt to ascertain its properties. When I used this technique on the HP switches, EotS successfully added the devices, properly reporting the majority of the devices' properties.
If that procedure had failed, or the resulting device information was unsatisfactory, Entuity's customer support would help me continue my quest for accurate data. Customers can also contact customer support to see if updated vendor device information files are available, which can be added easily. If Entuity hasn't created the vendor files for a particular piece of equipment, you can send them a text file, generated through an EotS SNMP-walk utility. Entuity would use that to create the necessary file, which it will then send back within 48 hours. Entuity compiles all the information it collects from customers to build additional device support into future releases. Customers can also download new device information from Entuity's Web site at any time. This solution seems fairly timely and easy to utilize, ensuring you can successfully update and correct the data reported by the EotS server.
Entuity support deserves a pat on the back for the EotS Bulletin Board, which shows any potential issues on the network. Its Bulletin Board details the events and allows quick access to information about the hottest problems. From each event listed on the Bulletin Board, I was able to access historical information about the node in distress in order to better determine if the problem was chronic or not (see screen shot). Standard threshold and SLA levels are given in a manner that makes it easy to compare them, though a quick jump to EotS's help site was necessary to fully understand the meanings of the threshold scales. Threshold settings are customizable as well, so you can have different tolerance levels for different nodes.
The UI was not without faults, though, such as sorting a port 10 above port 2 in a view. It's confusing to encounter nonsequential numbered items, and although it is common, it can be particularly annoying in this type of application. I was also annoyed to find I could rename objects or views in some windows, but not in others. The search feature in the component viewer also returned some interesting results on occasion, such as when searching for the IP address of the EotS server itself. I didn't always know which of my views came from which search hit, and I was often surprised by the location in the component viewer I would jump to when clicking on a result.
At present, the database schemas for fast ports (above 26 Mb, by Entuity's definition) and slow ports are different, which makes it impossible to create certain settings to slow ports, such as thresholds for port utilization. Entuity believes a 20-minute polling interval is too long to be useful for an alert, so it designed slow ports not to alert on utilization threshold violations. You can, however, re-categorize a slow port as a fast port, which can be polled every few minutes, thereby allowing for utilization threshold setting.
This, of course, creates a greater burden on the server and the network. I currently work on a network littered with 10Mbit, so I found this to be a considerable problem. Entuity says it's working to resolve this for a future release. Another stumbling block presented itself when the software did not inform me that I had exceeded my license capacity -- which impaired the software's ability to collect and report data. I had to manually run license checks on my own. Entuity is aware of this problem and is working on it.
EotS 2.6 has some bugs to work out, but it certainly gives other vendors a run for their money. Overall, I found it has a good UI, relatively decent discovery and good support. It is worth a look for both corporate enterprises and service providers in need of a stable, easy-to-operate network management tool.
RSS
