ManageEngine OpManager 7

OpManager monitors networked devices and the services and applications running on them. It runs on Windows 2000 Pro/Server, Windows XP Pro, Windows Server 2003 and Linux (Red Hat 7+ and Debian 3.0). The official system requirements are rather unusually specified – 200MB disk space and a rather intriguing 999MB RAM – but the usual story applies: the more memory and the faster the processor the better. The system uses a back-end database as its data repository; you have the choice at install time of using the bundled MySQL or pointing the system at your own SQL Server 2000/2005 installation.

Once installed, you use the system via a browser-based GUI. Because our test server already had a Web server running on port 80, the OpManager setup routine automatically decided to host its Web server on port 8060 instead, though this was a changeable option.

When you first connect to the GUI, you’re thrown into the “discovery” wizard which, as you’d expect, asks you for SNMP credentials and the address range in which you’d like to look for devices. It took a couple of minutes to explore our class C subnet and find the eight assorted devices, which included a couple of servers, a Windows desktop, a wireless access point, a print server and a wireless-to-Ethernet bridge. Rather disconcertingly the package drops stuff whose type it can’t figure out immediately into the “desktops” category. I’d much rather have an “Unknown” pot for new stuff (imagine the annoyance if you had 100 proper desktops being monitored and then ran an auto-discover only to find that it dropped a pile of new guff in your previously pristine “desktops” list).

For the devices the package doesn’t recognise, you’ll have to walk through the various unidentified items and define what type of device they are (firewall, server, etc), the vendor, and so on. At least you only have to do this once per device – which is lucky since the browser-based GUI makes it rather a chore. Incidentally, the system knows about 300-odd common devices, but it’s very simple to add new ones. Oh, and devices can at least be grouped together so that you can configure some parameters on several similar entities at once.

Being a monitoring system, OpManager can of course alert you when there’s a problem. You define how you wish to be alerted by defining “notification profiles”. There’s the usual choice of alerting mechanisms – email, SMS (either using email-to-SMS or an SMS modem), running an external program, or throwing a ticket into ServiceDesk Plus (another ManageEngine product that does trouble-ticketing).

The range of stuff you can actually monitor is pretty extensive. As well as the obvious items like checking that a device can be pinged and that a given service is answering (so to test an SMTP server it’d make a port 25 TCP connection), you can watch Windows services, event log entries and SNMP traps. For all these various monitor types you get a decent collection of built-in ones, but you can add your own with a few clicks and keypresses. There’s also a URL monitor, which is very handy for ensuring that, say, your Web server that lives in a remote data centre is continuing to answer requests. It’s worth mentioning, incidentally, that there are a few different versions of the product. At the bottom end is the free edition, which does basic monitoring of up to 10 devices. Next up is the Professional version – same features, but you can monitor more than 10 devices. Next is the Premium version, which adds monitoring for Active Directory, SQL Server and Exchange, and finally there’s the MSP version for data centres, which expands the system to a probe/server architecture.

The range of features in OpManager is very impressive, but sadly, the browser-based architecture really lets it down. I found the GUI sluggish, which made exercises like defining the make/model of the discovered devices extremely tedious. The developers have had a go with some nice dynamic HTML stuff - the very neat reports menu, for instance, but I can’t help feeling that a desktop-style architecture (either a native Windows application or a Java app) would free the developers – and therefore the users – from the constraints of an HTML interface.

In short, then: an excellent range of real-time dashboards and on-demand reports, very good abilities to monitor a wide range of devices, applications and services, and not a bad price either. But please, ManageEngine people, considering giving version 8 a proper GUI: you know it makes sense.