Home >  Articles > Tool turmoil

•  
•  
•  
•  

You can't improve what you can't measure. It's a simple mantra that explains the proliferation of management tools throughout modern business in recent years.

However, increasingly complicated enterprise information infrastructures are leading many companies to realise that even the best tools aren't delivering on their promise: management is often far more complex and error-prone than they ever believed possible.

Christmas in February?

Much of this complexity comes from the sheer number of things that must be monitored in today's environments: there are servers, desktops, storage arrays, telecommunications links, user authentication procedures, enterprise applications, network switches, virtual private networks, and more - and each often has its own management tool.

Consolidating the output from each of those systems into a single view has long been the purview of enterprise management platforms such as HP OpenView, Aprisma Spectrum, IBM Tivoli and Computer Associates Unicenter. Yet while they provide a common target for the introduction of new management tools covering specific technologies, the platforms are worlds unto themselves. They sport complex scripting environments and so many features that it takes a small army of technicians to make sure everything is running to plan.

With so many enterprise systems now interrelated, even a small glitch can set off a veritable Christmas tree of alarm lights. Unless staff are trained to sift through those error conditions, they can easily get distracted trying to fix problems that have little relevance to the real problem at hand.

Reconfiguring a specific server, for example, might improve performance marginally but it will do little to improve business service availability if there's congestion in the telecoms link. Firefighters, after all, aim at the base of a fire, not its flames.

And that, it's become clear, has proved to be an even bigger obstacle to effective enterprise management than the tools themselves. "It's easy to have a flashing light that says there's a problem with the PABX," says Oliver Descoeudres, Marketing Manager with systems integrator NetStar, which offers a menu of services including remote network management. "The most important thing is to have a critical mass of engineers or people who understand what's going on."

To keep from running in circles, companies need to learn the pressure points of their business and understand when the Law of Diminishing Returns has kicked in. That's driven many companies to take a more proactive approach to network management. Mercury Interactive, for example, has extended its focus in application testing tools into the enterprise with Topaz, a suite of tools that uses regular load testing and service-level reporting to highlight potential problems before they affect business performance.

"It's a move away from a device-centric view to a more service-centric or business-centric view of IT management," explains Peter Lilley, Field Marketing Manager with Mercury Interactive. "We don't want to micro-manage. If a device is humming along at 80 per cent efficiency there may be some room for improvement, but it may not be the fire we want to fight right now. A drill-down approach ensures that the problem resolution process doesn't create additional disruption, and that we're only solving problems that exist."

The buck stops there

When companies put in management systems themselves, the finer points of enterprise management can easily be lost on executives faced with ever-shrinking budgets. While a management initiative might be well-funded and well-staffed at first, constant pressure on resources will typically force consolidation within the enterprise management team. Still other staff members may be reallocated to other projects as they come up. The result, unless companies are careful, may be a management team that is better at calling for help to resolve problems than it is at actually doing anything about them.

The increasing demands of enterprise monitoring - fuelled by an overall trend towards focusing on business goals instead of technical esoterica - have driven many organisations to realise that network management is a challenge better faced by someone who lives, breathes, eats and sleeps the stuff.

Smelling potential millions in service revenues, there has been no shortage of contenders offering to do just that on behalf of their customers. Most major systems integrators now maintain dedicated network monitoring facilities, link into customer networks via a dedicated WAN connection, then use autodiscovery and a host of specialised tools to make sure everything's running to plan.

Australia has been particularly blessed when it comes to network management, since it's proved to be a popular location for multinational services companies wanting to provide follow-the-sun support to global customers. With 24-by-7 attention and a full roster of category experts at the helm, it's a service that has made a lot of sense for a growing number of businesses - particularly those that quantify the cost of downtime and don't like the number they come up with.

For TransAction Solutions, a provider of IT services to 3000 users across 11 credit unions, that cost was around $100,000 per hour. Recognising that involving a network management partner would help mitigate the risk of network failure, TransAction recently signed on for proactive network management services from Getronics, which watches the company's IT environment from its Belrose, Sydney network management centre.

"Customers just don't have the bandwidth to have people running and administering the tools," says Michael Jenkins, Network and Desktop Outsourcing Business Manager with Getronics, which monitors some 500,000 networked devices globally and serves 25,000 networked devices for 100,000 users in Australia alone. "We're all about energising those tool sets with tools, methodologies and higher-end skills wrapped around people."

Many Getronics customers, Jenkins says, are already running a management platform like HP OpenView but simply don't have the resources to use it effectively. "That leads to management being reactive within the organisation," he adds. "In our experience, I'd say that 70 per cent of network failures could be avoided if people were just looking at the right things and taking notice of the signs."

Taking the initiative

That's easier said than done, however. Whether handled in-house or outsourced to a specialist, keeping up with those signs can be a challenge akin to trying to drive through Rome without hitting anything. Signs are everywhere, but nobody pays any attention to them and you don't see where you were meant to turn until you're well past it.

With the mechanics of enterprise management now well and truly established - and every major platform now providing an abstraction layer linking networked components with the business processes they enable - management vendors are building a new generation of tools that proactively hunt down problems and resolve minor issues well before they become big ones.

"The tools are only as good as the processes you have behind them," says Karen James, Director of Operational Services with managed service provider Dimension Data. "Integration is critical," she continues. "It's really hard to have effective processes and people that are running a well-oiled environment if they're looking at eight or nine screens. You've got to have proactive tools to help out."

Concord Communications has carved out a sizeable niche in proactive monitoring with its eHealth suite, which monitors network, systems and application performance management and cross-correlates the three to spot potential problems before they happen. Getronics, Dimension Data and many other managed service providers have latched onto eHealth as a way of differentiating their service offerings from ordinary reactive alarm-watching. Proactive capabilities were, for example, a key reason that TransAction Solutions took the step to go with Getronics.

"The easiest thing for a company to do is to delegate authority for IT infrastructure to the IT guys," says Warren Ross, Concord's Manager for Distributed Application Management. "But companies need to think about performance with respect to their business processes. We understand how the infrastructure performs from the end-user perspective: it's in real time as the employee, partner, or customer clicks on the submit button or hits [Enter] when they do a transaction. We then integrate that performance measure with the performance of the IT infrastructure to understand what apps are slow and why."

Other vendors are taking proactivity even further. The latest version of IBM's Tivoli management platform, for example, was among the first of the IT giant's products to be fitted with what the company is loudly proclaiming as 'autonomic' computing technology.

Simply put, autonomic computing is the realisation of the common vision of self-healing computers. Thanks to established links between technical and business effect, this vision can be applied with great usefulness to the process of network management. Poorly running resources, for example, can be proactively identified and examined for potential configuration faults, resource conflicts, impending physical failure or any of a hundred other issues. Once a problem is identified, the system can recommend a solution or fix the problem itself.

The whole goal of this approach is to help humans focus more on the big fires and leave the small flames to the system. More importantly, by encapsulating business rules in technology, autonomic computing can help even technically-minded management staff prioritise their remediation efforts based on the areas that will have the most immediate benefit to key business processes. Without that guidance, they might easily waste time fixing the website when the EAI broker is the real culprit.

Andrew Belger, Manager for Enterprise Systems Management with IBM Tivoli, believes the introduction of autonomic technology will foster five different types of enterprise management:

• Basic (manual analysis and problem solving)
• Managed (with centralised tools but manual actions)
• Predictive (correlating resource behaviour to paint a wider picture of enterprise performance)
• Adaptive (where systems monitor resources, correlate results and take action)
• Autonomic (where tools automatically resolve issues according to defined policies)

"The way it's currently being managed is that there are lots of different tools for managing different pieces of the technology," says Belger. "Trying to understand that at a business level is very difficult, particularly for someone who doesn't have a technology background."

In the short term, management will remain a largely manual process. Analyst firm META Group believes that service level reporting will slowly gain intelligence until 2006, by which point the company believes what it calls "service-level 'everything'" will become partial reality. Interestingly, the firm believes that operational culture will be the biggest hindrance as automated technology improves.

Managing others' business

If management tool vendors have come a long way, spare a thought for their developers as web services grow in importance within enterprise applications. Whereas conventional performance management systems have at least been focused inside the four walls, web services will introduce massive variability in performance and function that could spell havoc for the business processes that depend on them.

Tracking these processes and benchmarking their performance will be hard enough - but it will be even harder to set up reporting structures so problems in third-party systems can be rectified quickly to minimise business interruption.

As autonomous management becomes more common, applications will begin to actively report on their performance to central consoles, saving vendors from having to find new ways to measure performance from the outside. This approach will be crucial to ensuring service levels across distributed applications (web services in particular) - something that Concord's Ross believes will be resolved as loose affiliations of partners build ecosystems of management accountability where problems are rapidly spotted and delegated.

"If part of a transaction went to a place where we couldn't put our management capabilities, we wouldn't know anything about it," Ross says. "But when businesses are dependent on each other, they'll have to help each other out," he believes. "Web services apps will share performance information. When you approach it from an application perspective, it really simplifies everything."

• Share this Article:
• digg this
• add to del.icio.us
• StumbleUpon
• RSS Feeds