Improving the effectiveness of your network alarm monitoring system
Blog: Capgemini CTO Blog
They are designed to view network operations in a single console to detect faults in the infrastructure. It is critical to detect problems quickly in any network so that network personnel can take corrective action. Failure to implement corrective actions may result into significant downtime, application failure, and, finally, loss of business – resulting in a poor customer experience.
The following statistics from the ITIC 2020 Global Server Hardware, Server OS Reliability Report highlight the importance of effective network alarm monitoring:
- An estimated 39 billion devices will be connected via the internet by the end of 2020. This is triple the number of internet-connected devices in 2015.
- 88% of most of the telecom network-based businesses require 99.99% reliability for mission-critical hardware, operating systems, and main line of business applications.
- One hour of downtime can cost up to USD300,000 which equates to USD4,998 per server/per minute
- Server security is a major issue. In 2019, the US experienced 1,244 data hacks exposing over 446.5 million records. Business email account compromise and phishing scams caused USD26 billion in losses from 2013 to 2019.
A typical modern enterprise comprises different network elements such as router, switches, etc. Modern IT requires the addition of new network devices and patch upgrades. The duration of these frequent changes to the network affects its performance adversely. A network alarm monitoring system typically generates different types of events. These events can be informational, sanity checks, planned maintenance, warning messages, trouble signs, critical faults, etc. Hence, it is necessary to detect, isolate, notify, and correct faults encountered in the network. This is called as fault management. The way these events are treated and analyzed ultimately determine how well the network infrastructure and applications operate. Here are some typical challenges that network administrators face when:
- How to effectively de-duplicate the incidents to filter out noise. Noise is attributed to undesirable and unproductive communication that results from overheads in troubleshooting duplicate incidents.
- How to develop a rule engine that will effectively cluster similar incidents to identify the root cause of the problem.
- How to predict a chain of events that might lead to severe damage or downtime.
- How to manage incidents across different geographies.
- How to enable automation to reduce incident management cost.
Rather than expecting the administrator to identify and classify large volume of events, it is necessary to use AI- and machine learning-based tools to accelerate the fault management process.
With modern fault management, an intelligent system can sort events and present only actionable faults for the administrator to work on. The system needs to:
- De-duplicate the incidents
- Isolate the fault event
- Cluster correlated events
- Identify the critical fault event
- Notify the administrator or in some cases resolve the event automatically.
Although there are products in the market with good features that provide a solution to manage network alarms, they lack in providing diagnostics for the fault events, are less effective in event clustering and require aThey also require some manual intervention, tend to have high licensing costs and they fail to point to the root cause events.
However, it’s been shown that effective event clustering leveraging cluster analysis to determine the root cause of an incident can prove beneficial. It is estimated that around USD370,000 can be saved per year by avoiding additional outages through effective events clustering. Operations staff can save 10 minutes per ticket and the ticket volume can be reduced by 30%.
AI/ML: They key to developing a successful network alarm monitoring system
With the rise of machine learning and artificial intelligence, telecom service providers can dramatically improve their approach to handling network maintenance alarms monitoring. With AI/ML, organizations can build robust systems to handle the network alarms to improve the maintenance response time and reduce costs. Additionally, algorithms can be trained on new data fed into the system, which means that, as more data is captured, actionable faults become more apparent. Additionally, AI/ML solutions can be implemented on premises or in the cloud and be configured to accept data from various sources.
Capgemini recently worked with a geographically dispersed telco organization offering to help it leverage AI for better managing the response to network alarms. As a result, the company boosted productivity by 20%, reduced costs by 60%, and witnessed significant improvements in the downstream incident management processes in terms of reduced manual effort for root cause analysis.
For more information please reach out to me or visit our website.