System shall speak unto system

Systems control and data governance have become contentious and complicated issues. Could an open source initiative hold the answer?

In the world of open systems, you might assume that supplier lock-in was a thing of the past. Openness should be a given, especially in systems and network management, which are all about managing disparate, multi-supplier infrastructures.

In reality, the opposite is true. The most popular infrastructure management tools are largely based on frameworks - underlying application programming interfaces (APIs) designed to enable different software products to talk to each other.

Frameworks are necessary because the larger systems management suppliers now offer so many different products that companies will rarely use them all. Instead, users pick and mix products, and the frameworks offer them a neat way of co-opting new products into their ecosystem.

Companies such as CA are moving past simple APIs towards integrated management databases that can hold data gathered by all management products in their portfolio, from storage management to security and service management. The idea is to make it easier for users to manage their systems from end to end.

This makes the IT systems better prepared for today's hot enterprise computing concepts, such as governance, performance management and lifecycle management.

Frameworks and management databases may ease certain aspects of enterprise computing, but there is a downside. Large systems management suppliers such as HP, IBM and CA sell integrated software suites that offer little in terms of inter-supplier interoperability.

Once you are locked into a supplier's infrastructure, you mostly have to use its software modules if you want them to work effectively with each other.

This kind of supplier lock-in is something William W Hurley II, chief technical officer at Qlusters, a supplier of open source systems management software, wants to ­address.

"On rare occasions, all three of the major systems management players work together, but usually each company wants their own standard to be adopted," he says. "We need to focus back on the user, with truly open standards that the entire community has a say in."

Qlusters is one of the founding members of the Open Management Consortium (OMC), a collection of open source systems and network management software suppliers that have clubbed together to try to make the systems management world more equitable.

The idea is to develop a standard set of interfaces for integrating different systems and network management products, explains Mark Hinkle, vice-president of business development at OMC co-founder Emu Software.

"The great thing about commercial software is that problem solvers were integrated into a broad solution," he says. "BMC, Tivoli et al all make stuff talk together."

However, Hinkle says that this creates a tension between integrated suites and point solutions. "We want to facilitate the conversation to get those best-of-breed applications to interact," he says.

Hinkle describes a situation where, for example, a provisioning system to set up a server might automatically register it with configuration management and monitoring systems.

Not only would the services and devices being managed be automatically registered on a central management console, but the monitoring system could instruct the configuration management system to take certain actions in the event of a problem, keeping the server hardware and software running without having to call in a human operative.

Such capabilities may already be available in single-supplier eco­systems, but Hinkle wants the OMC to help develop cross-supplier standards so that different products can interact to do the same thing.

Hinkle also hopes to bridge the proprietary and open source worlds, making it possible for this egalitarian framework to communicate with frameworks from the large proprietary suppliers.

He may have his work cut out. When asked whether CA would support the OMC's work, Jonathan Price, principal consultant at CA, defended his company's single-supplier framework model, arguing that it has benefits over integrating such technology.

"Our approach has the added benefit of providing inherent integration between the modules, so minimising the time to implement and realising the benefit earlier, rather than having silos of management within an organisation," Price says.

"With a point solution approach, customers bear the cost of the integration effort themselves. Every time product versions change, that integration has to be reworked - often a hidden cost."

Although some proprietary suppliers are unconvinced of the benefits of the OMC's aims, there may be other ways for the OMC to get inside proprietary frameworks. CA, for example, offers a CASmart certification programme to help technology integration partners hook their products into its framework.

In the meantime, the Oasis standards body approved IBM's Common Base Event (CBE), a format for describing events to systems and service management applications, as part of its Web Services Distributed Management model last year.

There are encouraging signs from other areas, too. Tom Bishop, chief technical officer at systems management supplier BMC, posted a message of support on the OMC blog, vowing to find ways to engage with members.

Bishop argues that larger, established suppliers and smaller, innovative companies need each other to drive the market. "We benefit from each other, and trying to posi­tion or spin this as competition is misleading and dangerous," he says.

We have seen this model before, where large and small suppliers co-operate around an egalitarian, open source framework, turning more mature technologies into commodities by standardising them and leaving room for more innovation at higher levels of the stack.

Eclipse, the project to set up an open source framework for the creation of integrated development tools, has gathered strength since its formation as a consortium in 2001.

It cast the more established elements of the traditional IDE market, such as editors for Java and HTML, as commodities openly available to tools builders. This left commercial tools developers to refocus on the higher end of the value chain, such as application lifecycle management.

Just as the IDE community has a stack (a set of functions that becomes more complex and closer to the business the further up the chain it goes), so does management software. Neil Walker, network management systems product marketing lead at Cisco, puts element management at the lower level of the stack (knowing the status of a Lan card in a PC, for example).

On top of that is traditional network management, where the top­ology of the network comes into play. Further up the stack, there is service management, where administrators are more interested in how applications and processes are functioning.

"As you move up the stack, the cost of integration gets higher. This is why we integrate our own products together," says Walker.

There are plenty of existing net­work management standards at the lower levels of the management stack. Simple Network Management Protocol (SNMP) has been used for years to query network devices, feeding data back to the network management system.

Pulling these into an open source-based commodity framework would be a good place to start, says Hinkle. "In the network space, all the participants have SNMP monitoring capabilities," he says. "Maybe we could take companies' agents and merge them so that you end up with an open source management agent."

Can the OMC achieve this by mimicking Eclipse's model?

"I am a huge fan of Eclipse," says Hurley. "We are trying to take principles from Eclipse and the Apache Software Foundation and apply them to systems management."

The difference between Eclipse and the OMC, though, is that the former was kickstarted by IBM, which essentially gave it to the market after founding and funding the consortium during the crucial first phases of its life. The OMC was founded by a loose collection of small companies.

Its members, in true open source fashion, emphasise the development of policies and standards through consensus rather than the use of a single, driving voice. Proprietary suppliers such as BMC may applaud politely from the sidelines, but the OMC has no IBM to drive it forwards.

That could be a problem, says Darren Meister, assistant professor of information systems at Canada's Richard Ivey School of Business, who focuses on the role of technology in organisational effectiveness. "Open source is not very good at big-picture planning," he says.

Developing standards by committee is also notoriously difficult, and Meister is not sure that the OMC has the industry muscle to make the market sit up and listen anyway.

But the Linux community has proved tenacious in the past, and its capacity for co-operation is well known. If it does work, then the next question is, to what extent? How far up the stack can a loosely coupled consortium like the OMC expect to get, and will we see its base framework embrace challenges like service management?

The industry muscle that Meister describes will only come from gaining new members. The danger is that as more members join, the consensual decision making model that it has adopted will fail to scale. The alternative is that a large company like BMC actively jumps on board and begins to steer the ship, but that would raise concerns over control and autonomy.

If the successes of open source operating systems, application servers and databases are anything to go by, open source infrastructure management products will soon take a more central place in corporate computing. The OMC is well placed to ride that current, but it will need a strong hand at the tiller to avoid drifting off course.

Vote for your IT greats

Who have been the most influential people in IT in the past 40 years? The greatest organisations? The best hardware and software technologies? As part of Computer Weekly’s 40th anniversary celebrations, we are asking our readers who and what has really made a difference?

Vote now at:

Read more on Integration software and middleware