Current Filter: Network>>>>>Opinion>

   

It is a well-established principle within ITIL and industry best practice that you simply must operate a proactive IT management if you are to have a sporting chance of beating the IT management odds. Without one you are driving blind, making changes and planning in your environment without vital data on which to base decisions.

But it's more complicated still, because relatively few IT management system implementations are successful. Typically most fail because of inadequate planning for the range of post installation optimisations that are required to complete a successful rollout, resulting in compromised systems that are incompletely installed, only partially functional, or that generate floods of meaningless alerts.

To guarantee success, first develop a policy that accounts for the workflow of the organisation and which establishes a realistic and attainable response plan to deal with incidents. Most importantly, select an IT management system that minimises the ongoing administrative burden that is required for operational effectiveness.

Begin this process by defining the scope of what must be monitored and then organise them into categories based on their operational importance. These may be internetwork links such as fibre or data circuits, devices like network infrastructure switches and routers, or applications - and by extension the systems required to deliver those applications correctly. One approach is to define the importance of these systems using the number of critical applications that they impact in the case of failure and then basing your prioritisations on that.

For each of the different priorities you've defined, determine the responsible parties and in what order and frequency they should be contacted. Getting notifications about systems that are about to fail is only useful if they can be repaired before an outage occurs, and this requires flexibility in the types and formats of the messages deployed. Getting a suite of useful information included in the initial message that alerts an engineer can reduce the time required to track down the problem. Define a reasonable escalation path in the event that someone can't be reached, or doesn't respond in a timely fashion.

Once you've defined what needs to be monitored and who to notify, implementing a system that can monitor as many of those critical systems and applications as possible, in a single platform, greatly simplifies ongoing operations and troubleshooting. Fewer places to go to establish what is happening reduces time to repair and eases the administration burden of keeping multiple systems synchronised. The management system you select should ideally be accessible from any platform, including mobile devices and tablets, as well as from remote client systems. If you are working from a remote office and none of those systems has the client software installed to talk to your management platform, you're in for a long day.

As part of the deployment and implementation of this system you should ensure that you are not over-configuring alarms. Many potentially harmful conditions are better handled with a daily or weekly report if they are not yet causing an outage. It is wonderful to be able to instantly know about everything in the environment, but being bombarded with alerts simply encourages engineers to ignore or filter them, and that can be worse than no alerts at all. At that point you are still relying on the users to tell you about outages.

Simply put: do not configure any alerts on the system that will not be acted upon immediately. Save everything else for scheduled reporting and evolve this strategy over time. If your chosen vendor can assist you through this process, it is wise to take advantage of their expertise and create the solution that your business really needs.