Environmental Hazards to Peak Performance: Why Network Monitoring is Important for System i Managers

Whilst no one can be aware of every potential problem that may or may not impact the system, IT Managers can reduce the incidence and impact of any potential problem in two ways. The first is to raise their awareness of the boundaries of their systems within the network and identify the areas of overlap where potential problems may occur. The second is to extend the proactive approach to systems management that keeps all their own elements running well, to these overlap areas, in an effort to protect the system from unforeseen problems and deal with them promptly and effectively when they do occur. Frustrated users, who are delayed in their work, have no means of pinpointing the issue, so often any ‘lag’ to productivity is initially deemed an issue for Operators to resolve. In the case where these may actually be network issues, a bitter blame game can ensue with both Operations and Network teams pitted against one another as they each dig up ‘proof’ that the problem is not theirs. As this lengthy elimination process is carried out, the users are still waiting and productivity and profitability still suffers. As an Operator, it may not be your fault that the IOP’s were incorrectly configured by an engineer, or that a user has sent a huge file that has had a knock-on effect on the network traffic, or the TCP Ping has fallen over and severed data communications between users and the server. None of it may be your fault – but with immediate visibility to these issues, solving these types of problems becomes a fast and pain free experience.

Calculating the Cost
Without sophisticated tracing tools to accurately pinpoint common ‘network’ issues that could be impacting users’ productivity, it is very difficult to know where to start looking in identifying potential causes. What’s more, as these issues have the potential to be either system or network related, quite often, members from both teams become involved in determining the root cause, effectively utilizing the time and resources of twice as many people. The costs of this type of system/network detective work can soon add up to significant sums. Frequently, issues such as this often fall into two major categories and should be primary considerations for anyone considering this type of network monitoring:
_ Network Bottlenecks and Errors
_ IOP Utilizaton

Case Study
Company x is a large retail operation that was struggling with network issues that were not immediately visible on their centrally managed System i network. The network supports 10,500 users nationwide and the company generates $2.1 billion in revenues annually. The IT Manager and Network Manager often wrestled with the same issues and decided to review the financial impact this had made in the past 12 months. The cost –impact results of the review were far higher than they anticipated, primarily due to delays and the associated cost of the time spent to identify the issue. The team was also acutely aware of the (as yet) unquantifiable fiscal damage to the brand reputation and loss of customer loyalty. The figures below only account for the immediate impact.

See No Evil; Encounter no Evil
When users report slow response times to critical applications it can be hard to determine where to begin checking to isolate the issue. This is especially true when dealing with a distributed network environment. For example, users in Belgium may be experiencing poor response times but between the data centre in the US and the users in Belgium, there are numerous components and possibilities as to the cause of the lag on the network.
In other situations, there may be ‘clues’, such as high CPU usage but again this only serves to highlight an additional consequence, rather than the problem itself. Without visibility as to the primary cause, operators and network managers face a long elimination process that could see them involving additional resources in the form of IBM server engineers to try to eliminate the possible reasons for the lag. A lack of visibility inevitably translates to a negative impact on the user community and as such, must be the primary goal of any network monitoring solution.