Network Computingwww.networkcomputing.com

International Network Services (INS) VitalSuite 3.0
VitalSuite from International Network Services focuses solely on application monitoring. We found it the easiest product to use and its architecture was the best we saw. At its core is VitalHelp for diagnosis, and VitalAnalysis for generating reports. INS also sells EnterprisePro Perfor-mance Monitor, which we didn't test; it offers SNMP management that com-petes with the network performance management functions in Apptitude's and NextPoint's products.

Print this Page E-mail this URL

Vital's key report, called Heat Chart, is the best way we saw to display the relationships between applications and measurements, users and application performance, and users and network segments. We easily produced charts to cross-reference any of the above groupings. These charts use icons to represent five levels of performance, from OK to critical. We really cruised with this highly condensed format, which provides a summary of performance statistics and SLA violations, when the cursor hovers over one of the pairs. When we wanted to see more about a particular set of application performance statistics, a double-click retrieved the specifics on how the summary was created--it's a great feature.

The number of canned reports is relatively small. However, we loved that we could link reports to see increasing detail, or to get a different point of view of a problem identified at the top level. For example, during our testing, a critical problem with DNS availability was flagged on the top-level Heat Chart. For a quick look, moving the mouse over the critically red icon showed that the SLA had been violated for over nine hours. Double-clicking produced a tabular DNS performance summary, which included throughput and availability by historical, peak, median, historical median, delta and SLA.

Graphs showing all of the same statistical categories over time with current, historical and SLA values can be summoned. Particularly noteworthy were the 90th, 75th and median values overlaid on the server load and DNS activity graphs. A look at the juxtaposition of the three made it easy to determine if the violations were anomalies, or experienced by most transactions.

VitalSuite can monitor an exhaustive list of transactions, but not all of our PeopleSoft transactions were supported out of the box. Using VitalSuite's Transact Toolkit, we defined the specifics of the transaction to the level of our choice. No rival touched VitalSuite for flexibility and ease of use.

INS offers a variety of agents with VitalSuite. The two passive agents can do the bulk of the work, and when they can't, the AutoMon active agent probably will do the trick. While we'd have appreciated the integration of the passive and active features into one agent, we found the division of labor acceptable. The AutoMon agent runs simple transactions such as DNS, TCP, SMPT and POP from locally or centrally stored scripts. Included are separate scripts for LAN-attached and dial-up connections. Vital Suite does an all-around excellent job of differentiating LAN and remote users. Yet we were disappointed we couldn't modify AutoMon scripts; products such as Ganymede's have a leg up here.

VitalSuite's architecture is top- notch. The VitalAnalysis and VitalHelp console are both completely Web-based; the performance is surprisingly good and the interface is easy to navigate. The system is easy to install and upgrade, and we were delighted with VitalSuite's ability to upgrade agent software automatically when the server software was upgraded. The server architecture is hierarchical, with a single master dictating configurations for subordinates within its realm.

Vital Suite is priced on a per-seat basis. If you are trying to monitor application performance across the network, then it is acceptable to place an agent per segment. However, if you need helpdesk diagnostics, each user will require an agent and the cost goes up. The more advanced of the passive agents costs five times as much as the low-end agent, as does the VitalAgent Automon. In practice, we would strictly use the advanced agents because they open a window in Windows 9x's networking performance. Luckily, we weren't paying for them!

VitalSuite 3.0, starts at $43,000, International Network Services, (888) 980-8844, (650) 318-1100; fax (318) 1600. www.ins.com

NextPoint Networks NextPoint S3
NextPoint says its approach bridges the gap between network services and network operations, real-time and historical analysis, and network and application performance. To a degree it is right. Its S3 came close to the head of the pack, but it just didn't provide the insights into application performance that VitalSuite did. S3 handles SNMP discovery, alerting, RMON, trending, service exceptions and monitoring, along with application monitoring. We cranked up the entire product, but focused on the application monitoring. S3's active desktop agent approach sends transactions across the network against live servers. Statistics are uploaded to a centralized server that employs an Oracle database. To generate reports and administer the system, the console is accessed from a browser supporting ActiveX.

While we very much liked S3 overall, we found some significant drawbacks. For example, we didn't like having to install and upgrade the agent software manually, particularly when agents resided on the other side of WAN links. We also thought that the console UI's performance flat-out sucked. Even after we turned off status polling, the seek time to read the database was frustratingly long. And S3's user interface left us cold.

At S3's core are intelligent baseline values used for reporting and exceptions. To set the exceptions, we accessed the browser-based ActiveX channel bar central to the user interface design. The functions are well-organized, but frustrating to access at times; the size of the database slows loading. We created groups for application monitoring, including those devices that made up an application service. These groups can be used to control user access when logged into the user interface.

S3 has excellent baseline functionality. Like FirstSense, EcoScope and Apptitude, it delivers reports on minimum, maximum and average baseline values. S3 adds another value called a signature, which is like a baseline, but takes into account what are normal minimums, maximums and averages. We found the signature to be a particularly good basis for setting exceptions, though we were able to set exceptions based on any of these values.

S3's application monitoring supports ODBC drivers for Oracle, as well as the commonly known TCP and UDP sockets.

S3's evaluation of performance is based on tests performed on network devices on server TCP ports and by monitoring the application. All three are set up independently, manually and painfully. Network monitoring is fairly straightforward, starting with utilization of network devices. By testing for services on TCP ports, and by defining specific application transactions, all three could be correlated to monitor the application completely.

Tracking of service levels or service management is well integrated. From a central server, we created synthetic transactions that ran from NextPoint agents against application and network servers. We created service performance metrics to compare against our measured data, setting thresholds that would define our SLA. SLA tracking is divided into meeting, marginal and exceeding categories for statistics such as total response time. These categorizations of total traffic easily will show transient failures and developing trends, without modifying the SLA. These service levels are then reported in the application matrix overview, where we could drill down into the specific data points for each part of the system being monitored. We found this a very nice touch that was obviously well-thought-out.

The biggest lapse we found in S3 is the distribution method for agents, which should be automated. The interface design is generally usable, but we found navigation far from intuitive much of the time. Our major complaint about the user interface is that it's not at all consistent. For example, a device will have a DNS name in one location, and an IP in another. The interface lacks menus to modify names and other parameters, sending you back and forth through different parts of the tool to make such modifications. We found it confusing and tedious.

NextPoint S3, starts at $12,000, NextPoint Networks, (978) 392-2026; fax (978) 392-2027. www.nextpoint.com