Plumbing the depthsPlumbing the depthsFrom NetworkComputing Magazine Vol 17 Issue 01 - January/February 2008 Increasing network dependency and complexity makes a persuasive case for knowing more about the inner workings of your network, says Christian Twardawa, CEO of Paessler AG Modern network architectures are incredibly resilient and if not self remedial, self compensating. So, it is easy for Network Managers to bathe in the perception that all is well. Often, self-compensating actually means that the network is working but with unacceptable efficiency and compensation. When eventually it stops working, rapid detection and remediation is business critical and without the necessary skills, tools and information, availability is threatened; prevention is better. Such a scenario really can be the tip of an iceberg and frankly the Network Manager will only be able to ignore it for so long. Users will soon suffer, if progressive and incremental degradations to network performance continue unchecked. The dominance of IP has unleashed a new wave of 'network applications' such as VoIP, video, and surveillance, each traversing the network. Real-time applications are less tolerant to the vagaries of packet loss, corruption and delay, adding up to one thing; you need to know what is happening inside your network. It isn't an option, rather a case of what you cannot afford to ignore… Creating an empirical view of what is normal for your network establishes a baseline against which to measure and comprehend variations. With a well connected understanding of business requirements, informed decisions can be made, as opposed to guesses, or even worse, ignorance. Some form of continuous background measurement in turn drives a second benefit, exception-management. For example, if you are alerted to substantial increases in network utilisation, it may well put you ahead of normal security notifications in detecting a Malware attack. So you see, measurement feeds a process allowing the Network Manager to analyse, act and prevent; in many cases before user impact. In selecting an SNMP based product to carry out this measurement, it is important to ensure that good real time graphical displays are available to monitor and trend devices and resources, such as routers, switches, memory and CPU. This principal extends to such commercial essentials as the web server. Establishing a datum for your network will not guarantee error free operation, so you need to consider how monitoring equipment can play a role in diagnosis and remediation. Without doubt and when combined with good basic network operating skills, packet level analysis can prove invaluable in identifying cause and directing resolution, returning control. With the current trend to 'Virtualise' network elements such as servers, it means that a new layer of protocol activity and complexity is moving in. This too cannot be ignored and it would be unacceptable to allow new problems to establish through ignorance and apathy. All too often, organisations regard investment in network monitoring as one with no return and with the dependency that most business now have on their network, this must be tantamount to negligence. In fact the ROI needn't factor, at least in getting started, as suppliers like Paessler provide Freeware editions of their products, allowing you to evaluate a tool before buying. You can spend a lot of money on such tools and this is perhaps where the bad reputation arises. Many offer lots of features and functionality you will never need; translating into premium prices, plus the hidden costs of training, installation, configuration and ownership. Your selected tool must be intuitive, easy to use and offer good value. It is not for business managers to make the case for investing in network monitoring, but the clear duty of the Network and IT managers to make the case, then deliver. The case, once made, should be enshrined in policies and procedures that protect the business use of the network, while providing good management data, including trending and performance reports that the board can understand and be reassured by. The objective remains; a reliable, optimised network… NC |