Although NetIQ may have changed the product name, it hasn't changed the architecture or its history of providing backward compatibility (and forward compatibility, for that matter) with its distributed agents, or "end points" in NetIQ-speak. We also use Chariot for testing in our partner lab at Schneider National, in Green Bay, Wis., so I wondered how this beta release, which includes version 4.1 end points, would affect Chariot, since both products use the same endpoints. Chariot 3.2 had no problems communicating with the version 4.1 end points.

Easy Distribution

Installation is a breeze. Although the end points are available on 29 supported platforms, the Configuration Manager module and Monitor Console require Microsoft Windows 9x or NT/2000. The good news is that End2End 4.0 also offers a smooth upgrade of end points via the End2End Configuration Manager. It's an excellent method of upgrading potentially thousands of agents automatically. If you prefer, the end points can be downloaded and installed from the End2End console Web site.

The End2End Configuration Manager offers many options, from the auto-matic discovery of end points to SLA (service-level agreement) configuration. Schedules for services, applications and network monitoring can be configured, and the thresholds for each can be set. SLA configuration is straightforward, though I was not thrilled with the limited choice of either throughput or response time for the basis of my SLA monitoring. I'd like the ability to monitor my SLAs from both perspectives. SLAs are tied to scripts or applications. You can define as many SLAs as you like, but thresholds are configured on a per-SLA basis.

End2End 4.0's interface (screen view) Click here to enlarge

Diagnostic offerings include the ability to run an application or service script on demand, with the click of a button, providing a "right now" view of network performance and offering real-time diagnostic and analytic capabilities. The application monitor watches traffic generated by users and then reports the response-time statistics for each transaction. Real-time monitoring of custom TCP-based applications can be accomplished by specifying the application port or port range.

Service scripting offers the ability to monitor "real live" applications via synthetic transactions. The supported services are limited and are not extensible by the customer. Scripts for new services will be added in subsequent releases of the product and end points, according to NetIQ. A transaction- recorder utility provides an easy mechanism to record your Web-based transactions and create a service script. The URL and any posted data are captured by the recorder and played back as synthetic transactions for monitoring.

End2End does not support interaction with Java and JavaScript. Any Java applets or JavaScript encountered during the run of a service script is ignored. If End2End is to continue its growth toward becoming a complete network and application-performance-monitoring solution, the support of applets and scripting in Web applications will definitely need to be provided.

After examining all the options, I set up schedules to run application scripts as well as two service scripts: one for my DNS server and one for a Web site. Then I let End2End run for a few days to see what it would discover.

Reporting

The End2End Monitor Console has gotten a face-lift. The Web-based GUI presents two new options--diagnostics and events--in addition to the monitoring, administrative, reporting and download (end points, scripts and performance data) options. The diagnostics tab provides the capability to perform tests on demand in real time for troubleshooting network or service issues. The events tab presents event tracking and severity information, as well as trend analysis.

The granularity of statistical data provided by End2End is excellent. Response and transaction times are broken up by DNS name lookup, initial connection time, redirection time, time to first byte and content download time for in-depth analysis of network performance. Reports for all the monitored types--services, application and network--show the number of attempted connections and successful connections, hourly breakdowns, and performance trends.

If you do not set specific thresholds for each script, End2End will compute these values for you based on previous performance results. The documentation indicates that End2End requires at least one day of data with at least 10 samples to compute this. However, you'll need to allow End2End to monitor the network for longer than that to be accurate--unless you consider a .001 second degradation of response time over a 48-hour period to be a downward trend in your network performance.

Reports are easily generated based on script type, and end point and server location, as well as traditional date based. Also handy is the ability to view performance statistics from a monitoring standpoint or by SLA. End2End also provides you with system monitoring data for its end points, offering a view of CPU utilization, disk I/O and memory paging statistics.

End2End offers SNMP support for traps and is distributed with its own custom MIBs. In addition, you can click a check box, and End2End will generate DDI files on a scheduled basis for integration with Concord Communications' Network Health 4.5 or generate data files for Trinagy's Trend. Of course, integration with NetIQ Operations Manager is included as well.

Although End2End is not a total solution--it still doesn't do inventory or analysis of design topologies--it's moving in the right direction. Its offering of exception reporting for application and network services is a boon, and its capability to drill down into collected performance data makes root-cause analysis of network and application-performance issues a bit easier to deal with.