NetQoS has put out new versions of its software and opened a UK branch to launch them into Europe.

Marketing VP Steve Harriman told us it was time for network managers to stop focusing first on network availability and instead place the emphasis on network performance and by extension, application performance.

"Fault and availability management is a problem that's largely been solved - routers and servers don't break often, even Microsoft operating systems don't break as often as they used to. Plus, when things do break, people have the tools to fix them," he said. "Performance management is a much more difficult challenge, and one that's not being accomplished as well as it might be. Often, the utilisation rate of the infrastructure is used as a proxy for performance - it's a very poor proxy though.

"The performance-first approach is that the best metric is to measure the response time of critical applications as the user sees them. Measure the round-trip time to the application - where is the latency occurring? How much is in the network, the server, the application? Having that visibility is a huge step forward, especially if it's done without agents, as we do it, because agents put a synthetic load on the network."

The biggest overall change in NetQoS Performance Centre version 3 is a shift to web services, said Steve Fulton, the vendor's senior director of strategic alliances. He said this would make it easier to integrate NetQoS tools with those from other developers. "We have built a web services architecture that allows very easy integration in and out of Performance Centre - it's as easy as clicking a graph to generate a URL and then pasting that into a page," he added.

He said that NetQoS has already used the technique to display data from Auditec's Newtest, which deploys software robots at key locations and generates synthetic transactions to test selected business processes.

NetQoS has also updated the three individual tools which it uses to track different aspects of network performance, and which feed data into the Performance Centre portal: SuperAgent and ReportAnalyzer, which were developed by NetQoS, and NetVoyant which it bought in.

SuperAgent monitors application performance without needing agent software on the servers. Instead, it polls the network traffic from a switch port in the data centre and works out from all that how fast your apps are responding and what performance your users are seeing.

In the latest version, SuperAgent's development has focused on permissions and roles (so you can view performance by user groups, say) and on usability. "Usability is often overlooked by vendors in our position, I think," said Fulton. "You will never operationalise these products if they're not intuitive. We want engineers, net-ops and managers all to be able to view data in ways that make sense to them."

The latest version of ReportAnalyzer, which tracks NetFlow and other flow data, can store more data; and NetVoyant does SNMP polling for infrastructure helath checks, and Cisco but can now report VoIP call quality too. NetQoS SuperAgent is £22,000, the other two tools are £14,000 each, and you get the Performance Centre console free with any one of them.