Wares extraordinaire

SuperAgent is a passive monitor, meaning it doesn't increase network load (a great feature in a performance tool). It also tracks responses interactively, meaning it checks performance multiple times during one TCP session. Other tools, in contrast, measure response time once, at the start of the session. The problem with that approach is it doesn't account for changes as time goes on; performance could degrade, and the tools wouldn't know why.

SuperAgent stores its results in a SQL database that can be accessed via a Web-based interface. It also will deliver reports and alarms via e-mail or populate an existing management system. Using NetQoS' Reporter-Analyzer, you can combine this information with information from outside sources, such as Cisco NetFlow, NetScout probes and NetIQ response-time agents.

When Network World tested this product earlier this year, it proved easy to install and highly accurate, and it delivered as promised on all points. I can't improve on tester Bob Currier's final words, "SuperAgent 2.0's combination of functionality, reliability and superb reports make it a definite 'must have.' "

Johnson is president and chief research officer at Nemertes Research, which provides quantitative research on the business impact of technology. She can be reached at johna@nemertes.com.

The Pick: iHateSpam Sunbelt Software By James Kobielus, Above the Cloud columnist Shielding e-mail from spam blasters.

Spam is the cosmic background radiation from the Internet's rapid expansion into our everyday lives. Spam is everywhere, coming from seemingly every direction. But that doesn't mean that I have to accept it or that I have to subject my e-mail system to a continuous, corrosive stream of incoming junk.

Spam threatens to debase e-mail as a communication medium, just as junk postings slowly rendered many Usenet groups uninhabitable. I like Sunbelt Software's innovative iHateSpam client software because it lets me totally shield my e-mail in-box from spam. At the same time, the product lets me inspect every message that comes my way, just to make sure it didn't block or discard any important e-mail. IHateSpam is certainly not the only client-based antispam tool on the market, but it is the only client-based product I've come across that builds "whitelists" by scanning your Microsoft Outlook e-mail address book and your Outlook client-side message folders. IHateSpam also is the only client-side antispam tool that quarantines any incoming mail that doesn't come from a whitelisted sender. Whitelist-based antispam filtering is the only effective way to ensure a continuously spam-free in-box. Another plus is that iHateSpam lets users dynamically adjust the client-side rules to ensure the tool doesn't mistakenly filter out legitimate bulk e-mails - such as e-mailed newsletters - that they regularly receive.

The product's name is none too delicate, but it succinctly expresses the user's primary motivation for seeking it out. IHateSpam comes in separate versions for Outlook and Outlook Express, both downloadable from Sunbelt's Web site. The product installs quickly and easily, and setup is straightforward.

Upon setup, the software creates a whitelist of accepted e-mail senders - also known as a "friends list" - by retrieving addresses from users' e-mail address books and by scanning existing client-side mail folders for the addresses of senders of stored e-mail. IHateSpam prompts the user to define an acceptable spam-filtering threshold, ranging from the stringent "all spam" (which also might quarantine non-spam such as online newsletters to which the user has subscribed) to the more permissive "some spam" (which will only catch junk mail with easily identifiable spam indicators). Users can define a personal "enemies list" of addresses from which incoming e-mail is always blocked, and more fine-grained personal filtering rules keyed to contents of message headers, subjects, bodies and attachments.

One of the most powerful and dynamic features of iHateSpam is the client software's participation in a peer-to-peer online "Learning Network" hosted by Sunbelt (similar to rival peer-to-peer antispam communities such as Cloudmark's SpamNet). Users can designate any message in an in-box or a quarantine mail folder as "is spam" or "not spam" by clicking on buttons embedded in the mail client. When a user designates a particular message as spam, the iHateSpam client software sends a pattern of that message to an automated rules-creation engine in the Learning Network. The engine tabulates "votes" from all deployed iHateSpam clients, ranks messages by the probability of their being spam, and automatically creates new spam patterns and filtering rules. The engine regularly distributes updated spam patterns and filtering rules to all iHateSpam clients. As a final check, Sunbelt personnel review all patterns and rules before their distribution, just to make sure they're accurate and that user cliques aren't unfairly ganging up on legitimate commercial expressions (such as a competitor's e-mail-based direct-marketing pitches).

A big caveat for iHateSpam is - and for any antispam solution - that it relies on client-based rules-filtering. The product consumes CPU cycles on client machines, and briefly freezes the mail client while scanning, filtering and handling incoming mail. This computational load can be a significant burden on low-end workstations and can prove disruptive for users who receive a lot of e-mail, especially as the set of spam patterns and filtering rules grows. Ideally, much of the filtering should be done on a centralized basis, by mail gateways hosted internally by an enterprise customer or externally by Sunbelt. However, the vendor has not decided whether to evolve iHateSpam in that direction. Clearly, the product will hit a scalability wall soon, might not be appropriate for broad corporate adoption and can't address spam-based distributed denial-of-service threats if it doesn't centralize some CPU-intensive mail-filtering processes. Nevertheless, Sunbelt's iHateSpam delivers clear value where it counts: by enabling users to keep spam from constantly interrupting their lives and cluttering up their in-boxes.

Kobielus is an Alexandria, Va., senior analyst with Burton Group. Reach him at jkobielus@burtongroup.com.

The Pick: Virtela VPN Virtela Communications By Steve Taylor, Packet Evangelist columnist VPN service with a twist.

Innovation comes in many forms. Truly unique ideas are rare. Less rare, and generally more productive, is innovation that comes in the form of taking a couple of old ideas and combining them in a unique fashion. This latter type of innovation makes Virtela Communications' Virtela VPN my choice as a category-breaker.

Several givens are shaping the long-haul services market today. Companies are demanding extremely reliable services. Stability of individual facilities-based service providers is a primary concern. The price has to be right. And most of the network traffic is evolving to - if it isn't already - IP.

Add to this a couple of industry-related factors. On the services side, most facilities-based carriers have vastly overbuilt their IP backbones, and the majority of the Internet backbone is, in reality, lightly utilized. Still, users are concerned about the lack of control and quality of service (QoS) in the Internet, especially if their traffic is primarily handled by one ISP.

On the equipment side, several vendors now offer IP service switches that provide tunnel management and measure QoS across IP backbones. These switches let service providers control a large number of secure tunnels and to impose a number of QoS parameters across the otherwise less-than-ideal IP backbone.

Throw in the reality that most service providers have historically bought and bartered bandwidth from other service providers - the carriers' carriers. This long-established, but not widely publicized, practice lets carriers provide bandwidth to each other over physical fiber or large dedicated bandwidth pipes that is subdivided and resold to corporations.

Virtela has taken advantage of the best of each of these factors to create a next-generation VPN service that provides QoS. It doesn't own its core network. Instead, it buys otherwise unused Internet bandwidth in bulk from facilities-based carriers. Then it puts its own IP services switches on the edge of the network so it can add QoS and secure tunnel management on the outside.

This multicarrier approach brings three fundamental benefits (see story, "Top tips for skating through the telecom meltdown"). First, because Virtela is buying bandwidth without the network infrastructure, it can offer the service at an attractive price. Second, the service switches let Virtela superimpose QoS with some strict service-level guarantees. This is possible because if one network is not providing the desired service level, Virtela simply can switch the traffic to a different part of the Internet backbone. Third, for those who are concerned about the future and stability of individual service providers, the multiprovider approach gives several back-up paths, insulating the customer from a single-carrier network failure. Virtela, which is one of Network World's 2002 10 start-ups to watch, offers a number of enhanced services, including turnkey voice and video, on top of this VPN backbone. But the company's unique infrastructure and market approach really provide the power behind the service.

Taylor is president of Distributed Network Associates in Greensboro, N.C., and publisher of Webtorials.com. He can be reached at taylor@webtorials.com.