Monitoring the health of enterprise applications: What’s wrong and how to get it right

The digital economy is all about conducting personal and business transactions over internet powered applications that make it simple, fast, efficient, and cost-effective. Be it personal banking, processing payments settlements, buying stocks on an exchange, calling for a cab, ordering food, application for government tenders, storing land records, dating, registering marriage, or paying utility bills, these applications have made an impact on every aspect of human lives. The sheer volume of these transactions has posed a challenge to the large technology enterprises managing the internal and external business applications. One glitch in any component could bring the entire system down and cause significant loss of revenue.

To avoid such hazardous situations, it is critical to ensure the monitoring of every component of the application stack round-the-clock. However, the traditional approach for monitoring has some flaws such as:

Partial view could be blinder: IT teams have always been following a fragmented approach for application monitoring rather than using a unified view. They use hosted or SAAS based monitoring solutions designed to monitor specific application or setups. As a result, a single enterprise could have multiple solutions to cover the full stack. This approach is neither efficient nor effective. Instead, a better way could be to use a centralized monitoring solution that visualizes the application as a whole and not sums of parts. Such a monitoring system provides infrastructure monitoring, application monitoring and sends alerts when a predefined metric breaches the threshold. This monitoring system can also reveal valuable insights about the system which would have never been known with scattered data sets.

Static KPIs are dangerous: Often IT teams to set up monitoring KPIs for the application and just forget it. This is almost suicidal. As the application evolves to accommodate new data sets, provide new services, and deals with new threats, the monitoring KPIs should change in line. In other words, KPIs for monitoring the application stack should be a dynamic endeavor. Even though some monitoring solutions provide plug-ins and customization features, the IT team still find it challenging to monitor the KPIs which are critical for business decision making.

Weed out: Often, large scale enterprise applications generate a huge amount of data, both relevant and irrelevant. Unless the monitoring system is customized based on the most relevant KPIs, the system will continue to be clogged with irrelevant monitoring data which misses the key KPIs.

To deal with these flaws, the new age enterprises need to move to a monitoring solution which is:

• Adaptable: Every enterprise aspires to have a unique business model and cater to different use cases. However, the monitoring solution needs to adapt to these models, use cases, and be apt for on-premise or cloud deployment. If the system is built using customizable open source applications, it will offer this adaptability. The admin or DevOps teams can easily write custom programs which collect mission-critical data. It should also generate alerts to detect unusual behaviors in any segment of the application and take the necessary actions. Based on big data analytics and machine learning capabilities, the system can analyze historical trends and help in capacity planning. From the reporting point of view, it should produce shareable customized business specific dashboards. These dashboards could be further customized for non-technical functions such as sales and marketing. There could be several valuable insights from these dashboards based on heaps of data available on application servers.
  • Managed Solution: We cannot leave too much to automation, especially when it comes to applications which form the core of businesses. Having a managed solution for application monitoring implies that the system is always under additional manual support and offers patching and software updates.
    • Safe and secure: If there is not enough said and written already about the safety and security of enterprise applications, here is a chance to add some more. The monitoring solution should provide role-based access to all the important metrics of application, without loosening the grip on server level access. This means that no bit of data leaves the safe periphery. 

Narinder Kumar

Guest Author Narinder Kumar is the EVP, Technology Services at TO THE NEW. He is a strong proponent of Agile methodologies and is a Certified Scrum Master(CSM), Product Owner (CSPO), Certified Trainer for CSD (Certified Scrum Developer), Cloudera certified Apache Hadoop Developer and Data Analyst. He was also a co-founder of Inphina Technologies.