The digital operations of India’s largest media firm Hindustan Times (HT) have achieved a marked improvement in its core performance metrics by deploying an observability platform that collects and contextualises operational data from various sources.
The solution, from New Relic, helps HT’s digital products provide a consistent customer experience.
“We improved core performance metrics from ‘poor’ to ‘excellent’ by achieving true observability with a ‘single pane of glass’ view," Hindustan Times Digital Streams chief technology officer Shubham Srivastava said.
Srivastava added that New Relic has helped the nearly century-old media organisation achieve full-stack visibility of its digital operations with a real-time, in-depth view of infrastructure, network and applications.
It has fine-tuned its core software platforms ensuring high availability, low latency and near hundred percent uptime.
Srivastava said the engineering team had earlier implemented multiple monitoring platforms to ensure the deep integrity of their applications.
“Using heavily distributed and scaled NoSQL systems from the likes of MongoDB, Cassandra [and] CouchDB databases, [and] fast-moving index caching systems like Elastic, we wanted a comprehensive observability platform that can help to monitor everything from a single dashboard,” he said.
With a focus on managing larger traffic volumes and creating operationally centralised key performance indicators (KPIs), HT created a complete unified single dashboard for all different platforms, applications and software architectures.
“It is now possible to instrument, visualise, analyse, and troubleshoot our technology platforms running on a real-time cloud-based distributed environment,” he said.
“We have improved KPIs both in terms of the business and engineering”, he added.
Srivastava said some of the KPIs are page experience, core web vitals, mean time to repair (MTTR), user engagement, and a count of support tickets, all of which led to a better consistent product experience.
“Results showed good improvement in core web vitals, throughput, low latency in APIs served through backend applications and also improved product experience across low bandwidth, unpredictable network conditions. All of this helped to improvise a couple of relevant technical SEO metrics which would improve our organic user acquisition,” he said.
Srivastava pointed out that the organisation initially used a mix of processes and events to measure a certain set of data points around performance, and observability on a manual basis or through some open-source observability platforms.
“While these methods helped to provide tactical visibility into operational issues, with New Relic we could obtain more critical comprehensive real data insights across different sets of platforms and layers in our software stacks.”
The insights are able to expose the root cause faster while correlating infrastructure health with performance and customer impact on one platform, he added.
“Moreover, the observatory platform could penetrate across a portfolio of properties to figure out troubles in a consolidated distributed working system, thus enhancing the abilities of tech teams to track, reduce and resolve issues faster,” Srivastava added.
With New Relic query language (NRQL) and features like drag and drop, developers can write custom queries, build dashboards, and write alerts all in one place, he said.
However, Srivastava said it was not easy to implement New Relic custom integrations at the initial stage and developers required a lot of upskilling and perseverance in taking it through in production.
“Also given that we use various distributed complex components, getting to a unified view with all integration sorted was expected to be demanding,” he said.
Srivastava said firms have to have clearly defined KPIs, which should be made visible through the observation platform like New Relic.
“For mature digital organisations, insights and optimisations sourced through an observability platform should become an integral part of the SDLC [software development lifecycle],” he added.