Product Reviews

Reach Out and Watch Your Network

Longitude's agent-less architecture gives you a flexible solution for network monitoring.

• By Rick A. Butler
• 02/01/2007

There are certainly pros and cons to both the agent-based and agent-less approach, but with all the interfaces out there like WMI, it amazes me that some products today still mess around with agents -- but that's just me. Longitude from Heroix immediately stood out from the rest of the pack, in my opinion, as there are no agents to install on the managed nodes. After breathing a huge sigh of relief, I dug around in the tool and found it makes for a great enterprise-level monitoring solution. Did I mention that it's agent-less? I think you get the point, so let's move on.

The Management Station is the central point of activity in Longitude. The entire platform is Web-driven: a Java-based application running on top of a Tomcat Web service. This means you can manage Longitude from anywhere just by contacting the Management Station though the browser, anywhere on your network or the Internet, if it's properly configured.

Secret Agent-less Man
The Management Station can passively connect to a number of operating systems and application platforms, such as Windows 2000, 2003, XP, Linux, HP-UX, SunOS, AIX, Oracle, SQL Server, MySQL, WebSphere, WebLogic, Apache, IIS and Exchange Server. One thing I love about Longitude's approach is that it talks to systems over standard protocols and interfaces, including HTTP/HTTPS, WMI, SSH, REXEC, TELNET, JDBC, JMX and SNMP. In most cases, these interfaces are either available by default on the systems you're going to be managing, or it's not a big deal to install them. The point is that Longitude sticks to industry standards, which I certainly appreciate.

Now, there is an agent in the box, but it's not for installing on the end nodes. It works more like a DHCP relay agent or a bridgehead node. The Management Station can direct the agent to collect statistical data from the remote managed systems and bring it back to the Management Station.

This method works well when crossing domain or platform barriers. You can configure security settings as tightly as necessary and still allow the appropriate level of data gathering. This is a great feature, as it lets you shape the data stream leading back to the Management Console in a more manageable fashion. When every machine reports directly to the Management Station, it can create a lot of parallel traffic over your network.

[Click on image for larger view.]

Figure 1. Color-coded displays give you a quick glance at factors like CPU performance and service availability.

Installing Longitude was almost obscenely easy. Just install the tool on your system and within a few short moments, you're looking at the Longitude interface in your browser window. It's easy to get going with Longitude, another aspect that improves its positioning among similar tools. With many utilities, you'll have to spend time installing the tool and then rolling out the agents to manage your target population. With Longitude, you're done in about 10 minutes.

One cool feature that's not necessarily new -- but is always worthy of mention -- is correlated events. Essentially, you can set Longitude to flag an event based on a mixture of different event conditions. You might say, for example, "Alert me if Server 1's CPU exceeds 90 percent and disk space falls below five percent availability." You may not care if the processor goes over 90 percent by itself, but if that disk goes below five percent availability while the CPU is pegged you can have the system fire off an alert. To do so, you'd configure Event Conditions and tie them to your correlated event.

You can also use correlated events to customize reporting. For example, you can have the correlated event say that your project management group can't work because their Web server carrying the PMIS is offline. You create the defined event and use the correlated event to shape it as needed.

Longitude lets you monitor events to keep you in line with your Service Level Agreements (SLAs). For example, if you agree that the database server will be available 99.9 percent of the time with only one percent of degraded performance, you can set the criteria for what is considered good and degraded performance. Longitude will happily trend to see if you're meeting that requirement. This is nice because it takes the subjectivity out of the equation and gives you hard data by which to act on whether or not you're meeting your SLAs.

Monitor More
Stepping up from version 3.0 to 4.0, Heroix added more infrastructure monitoring elements to Longitude. This means you can monitor things like Cisco routers, DHCP servers, Active Directory, HP Insight and Dell OpenManage. Longitude has a broad array of operating systems and applications it can manage. It seems the folks at Heroix are always adding to that list, given that the tool first rolled out in the Spring of 2005 and it's already up to version 4.0.

A centerpiece to the new version of Longitude is the consolidation of event logs. A big concern for IT admins is mining through event logs to locate specific incidents. It's a unique pain when you have to do that across numerous machines.

Longitude makes it easy to work through event log monitoring. It creates filters you can use to sort through irrelevant events and focus on the ones you need to see, regardless of the machine or platform on which the events happened. Having events from a Unix box alongside Windows events is a nice feature and truly unifies your system management efforts.

The only gotcha I could put a finger on with this tool is that it does like to have a bit more RAM and processor on the Management Station. It needs 1GB to install in production mode and a 2.4GHz processor is recommended. Nowadays, though, with hardware priced as it is, this really should be a non-issue for a current IT platform. Just know you can't skimp on the machine you choose to run the Management Station.

Longitude relies on constant network connectivity to properly gather and report on system data. The only real drawback to a passive monitor is that if something prevents the machine from being monitored, like a network interruption, it will directly affect the statistical data being fed into Longitude. That being the case, Longitude probably isn't as appropriate for monitoring across remote or temporary connections. Since it doesn't use end-node agents, it can't collect data unless the Management Station can see the node. You can tweak the sampling times in Longitude, but all things created equal, you can't monitor what you can't see.

[Click on image for larger view.]

Figure 2. The statistics dashboard shows performance data with several types of graphs.

Longitude is equipped with a nice array of reporting structures out of the box. You can run reports on an ad-hoc basis or on a schedule for specific applications, computers or departments based on how you've tagged the machines. You can create dashboards that let you watch what's happening in near real time (the minimum is 10 seconds). I did notice, however, that this level of monitoring is highly performance-dependent, so if you're short on resources, this might not function quite as well.

The Verdict
The fact that Heroix Longitude is agent-less is huge for me. In my opinion, that's how a monitoring tool should operate. By not using agents, the monitored machines aren't directly affected by the performance of the monitoring software.

The ability to create correlated events and reports and align to established SLAs lets you use Longitude to map to business expectations quite effectively. This level of visibility is available with very little intrusion or investment of time, which makes Longitude well suited for anyone needing a no-nonsense monitor.