Hewlett-Packard will port OpenView Operations (OVO) to Linux, and is renewing OpenView for Unix.
The first Linux-based OpenView modules, which are due next year, will target applications where Linux is most popular, such as security management and the finance industry.
OpenView can already manage Linux systems, but the management console itself must run on either HP/UX, Solaris or Windows, said Willi Klenk, HP's research and development manager for OVO.
He added that demand for a Linux version mirrors what happened with Windows.
"What we are seeing with Linux in the market is a kind of deja vu," he said. "Like Windows, it has its favourite application areas - file and print servers, database servers and web servers."
The differences between Linux and Unix mean that a straight port is not feasible. However, Klenk said that development of OVO for Linux should be easier than the port to Windows, and will also be used to rejuvenate some of the code in the Unix version which is around 10 years old.
"OVO for Windows was mostly a rewrite. Linux is closer to Unix than Windows is, but it still has different protocols and APIs," he said, adding that he expected Linux to become the primary management platform for OpenView in the long run.
"Linux will go into specialist applications first, it won't be replacing HP/UX or Solaris for enterprise management next year, but I believe it's going to happen. It is difficult to predict when - after all, it has taken seven years for Windows to be seen as a reliable enterprise platform."
HP is also working on agent-less management. Most systems under management today require a software agent that communicates with OpenView, but this will be less common in the future.
"We are integrating two elements into our portfolio," he said. "One is probing, where you just ping the system - it's a possible way to reduce the cost of implementation if you don't need in-depth monitoring.
"The other is that operating systems are, more and more, providing similar functions to our agents, exposing a lot of data through standard interfaces, like an embedded agent. It does have limitations though. You can't get to all types of data, and it can create scalability problems with the amount of network traffic it generates."
Bryan Betts writes for Techworld.com