With voice and data converging, and the increasing importance of Web services, it's time for network management to give way to universal performance management, says Michael Szabados, the CTO of software developer NetScout. In this new environment, the network manager will need to take on new roles where the key is their ability to gather data and collaborate.
"Network management is the predecessor of performance management, which embraces a lot of other functions, especially importing and sharing information," he says. "Network management was a Lone Ranger deal, performance management is a team activity where you share problems and resolutions to get the implications."
Performance management is a much used - and abused - term in the IT industry, with a whole raft of software developers claiming that their products can improve the performance of applications, database systems and so on. Szabados sets his sights higher though, seeing today's network management as the core of something rather more complex.
"We want to be a universal performance management tool, integrating applications for troubleshooting, monitoring, event tracking, and so on," he says. "It's much more practical and feasible for an organisation to have a person versed in one tool than four - it cuts the clutter."
However, as software tools become more capable and complex, the challenge facing their developers is to prevent the complexity from feeding back and making that four-in-one software four times as hard to learn and use.
"A lot of our effort has been to enhance usability, for example more usable reports and functions," Szabados says. "The other key direction is to embrace more of the data sources in the network, and the Holy Grail is to connect it with business services - our strength is providing the enterprise picture."
Start at the data store
"We believe that to solve business goals, you have to start at the data store. You have to recognise applications and URLs, measure response times, recognise remote sites, and so on. If the data is not there, it doesn't matter how good your reporting is. NetScout's probe is the most powerful data source - it gives the link between applications and the business world and the network."
He argues that to complain about the high cost of network management software, as Techworld readers did in our recent survey, is to miss the point. This stuff is expensive because it needs to be, he says, and the organisations that need it - which tend to be large enterprises with big budgets, in his case - know that.
"In our price zone there's no measurable elasticity," he adds, pointing out that if NetScout did cut its prices or do a cheaper version for mid-sized customers, which could still happen, it would also need to change its sales methods. However, he argues that it is not suited to the pile 'em high, sell 'em cheap strategy, because where network management is needed, it is needed in spades.
"It's the same problem as LANs 20 years ago - the channel is still coming off the high of selling big iron and now margins are razor-thin," he says. "But network management is not quite analogous to LANs because it is exponentially valuable as the size of the network grows. Smaller networks have fewer problems, fewer applications, fewer users and less complexity."
Convergence and Web services
Looking forward, Szabados highlights a number of areas that will increase the network (or performance) management load over the next few years. "Voice, and the social issues of different voice and data groups, has been the radical break in the last two quarters," he says. "It impacts both midrange and the mainframe - in fact, the midrange started on voice earlier."
He adds that as networks converge and as IT functions connect to business services, those social problems of different priorities, usage patterns and knowledge bases can only grow.
"Social aspects have always been important, with administrators and application developers not understanding each other," he says. "So you always focus on providing data to close that gap. It's a problem for Cisco because it doesn't have empathy with the voice people."
One other growing area that he advises network managers to get their heads around is Web services and the changes they will bring to usage patterns.
"Web services will be a big driver, they're still in the hype phase now but they're real in substance. We're seeing a huge IBM commitment in WebSphere, for example - we're working with IBM on that. It creates a new level of complexity, sure, but it solves another set of problems, which is it addresses the access problems."