Categories
Blog

Using Network Automation for Health-checks

Often, network administrators spend hours and hours on troubleshooting on daily basis. When you have a complex network setup with large number of service provider links with dynamic routing protocols, things get even worst.

Why is it so common to find daily break-downs or poor performance of networks? And its quite evident that the time spent by network administrators to troubleshoot daily issues is not only a waste of valuable time but also a contributor towards frustration and stress.

Figure 1: Simplifying troubleshooting with automation

A network is a manually configured set of rules that should ideally function as planned without giving errors. However, this is not the situation as most of the times it fails to function as planned, resulting with a degraded end user experience. Why does it happen so often? Network is an interconnected set of nodes, that communicate with each other on protocols to provide traffic patterns to cater business needs. Different protocols exist, and they operate on different layers of the TCP/IP stack. Changes in one node could result with changes taking place at multiple other nodes within a network at no time. This is worse when you have a large network that connects through service provider links that has various dependencies on service provider network which is merely a cloud from a point of view of an enterprise. These changes could be minor changes such as port error, low memory or major changes such as an outage. Either way a change in one node can result in different traffic patterns based on path selection, low response time, low quality or non-responsiveness.

There are two ways that a network administrator can visualize changes on a network or on a node. One is via SNMP alerts and the other method is the tedious task of checking changes one by one on the node by logging in to the device. Although SNMP give network administrators some level of insights such as high memory, CPU and bandwidth details which are based out of MIBs, it lacks insights of changes on a node ‘state’. Identifying network state is part of troubleshooting and this is where most network administrators spend their time on. Manually going through line by line on a node to detect any changes.

Since this is done manually, its’ not a systematic approach to troubleshooting nor there is a standard process followed by administrators to troubleshoot a given incident. Further, since detection of errors or changes on nodes are observed manually, time taken to ‘detect’ is high with a lower accuracy ratio. When the node count is large, and interrelationships of nodes are complex, it makes a tedious task to observe deviations in each node and to identify patterns of failures within a network. This makes troubleshooting and root cause identification a time-consuming job.

Several measures are taken to resolve a ticket upon the detection of an error or a deviation. One of the measurements is to revert the node back to a previous ‘working state’ to see if things work fine. However, probability of knowing what the previous ‘working state’ is very low as there is no track of the dynamic state changes of nodes. Therefore, trial and error are used as a measurement to revert which is again a less accurate approach to problem resolving.

Once a potential resolution is found, next task is to test if it works for multiple scenarios. Testing is also done manually which is again a time-consuming task that often result with lower accuracy.

What does it all make? All this make daily routines extremely tedious and stressful consuming valuable time of network administrators, minimize network uptime and impacts end user experience.

How does automation help to resolve these challenges? Automation brings efficiency as repeatable tasks are performed by a machine systematically at a much higher speed than a human would do with greater accuracy. It will give you a regular view of deviations of network ‘states’ helping you identify issues and interrelationships of deviations to isolate a root cause. Testing is automated which means it’s much easier to verify multiple use cases at a very short time. All this makes higher network uptime, peace of mind and a superior user experience. 

Leave a Reply

Your email address will not be published. Required fields are marked *