Gaana.com: Serving billions of API requests a day

Music streaming service Gaana.com has become a staple among Indians listening to a wide variety of songs. CEO Prashan Agarwal explains how the platform allows greater flexibility to offer a more personalised music streaming experience to India’s mobile-first internet user-base.

This decade has seen both the smartphone and data explosion. Tell us about your journey in terms of that?

Gaana was launched in April 2010 and we have steadily secured market dominance over the last decade. We surpassed 100 million monthly active users earlier this year. Looking back at the past, the launch of affordable 4G plans was surely one of the biggest turning points for music streaming in India, as the market has exploded to more than 7X its size within 3 years.

Thanks to our deep obsession to better understand consumer preferences, we have been able to be the primary beneficiaries of the mobile internet wave, and have since launched an array of innovative products that have struck a chord with a wide section of first-time internet users across urban and rural India.

What are the tools you use to manage your songs and the traffic? Is the traffic consistent?

We use Hive, Flume, Hystrix, Spark, HBase, Vertica, Mongo and ELK Stack regularly to manage tech operations at Gaana. Our music library is hosted on Times Internet’s cloud, and consumes around 600 terabytes of storage. We rely heavily on multi CDN to serve and stream this content to our users. Gaana’s music player is heavily customized to play adaptive bitrates based on a user’s network bandwidth. Our consistent investment into tech infrastructure helps support the rapid growth in our user-base. In terms of music consumption, the traffic on Gaana is consistent throughout the week, with late nights and early mornings counting as off peak hours in general.

What about the backend? What about Big Data and your usage of data analytics?

Gaana’s cloud infrastructure is powered by approximately 700 VMs, and we serve billions of API requests every day, with our core APIs having an average response time of 1.2 milliseconds only. Our backend is mostly on micro services architecture which is designed to be highly resilient and robust. Our infrastructure can scale horizontally to accommodate rapid business growth. In terms of adopting emerging technology, big data and analytics are core to decision making in Gaana—be it user-related, product-related or business-related. We have tuned our search core specifically for Bollywood and regional music to serve best-in-class music recommendations as we have conducted extensive research on how Indians search for music online.

Tell us about your mobile app and how it has evolved down the years.

Our country’s music streaming space has evolved rapidly—both in terms of content consumption as well as platform innovation. As one of the early movers in the segment, we have been pioneers in drawing potential users in favour of using our app over side-loading music from questionable sources.

Today, while all major functionalities of our platform like searching for music, browsing and listening to songs among others are available on both app and web, some of our more recent innovations are mobile-only as the platform allows us greater flexibility to offer a more personalised music streaming experience to our country's mobile-first internet user-base.

Your app also has the voice assistant feature. What is it about and are you doing any research in Artificial Intelligence?

At Gaana, we have launched several industry-first features—be it making the app interface live in 12 languages, being the first music app to introduce in-app Voice Assistant, introducing sing-along lyrics on the player page, or Smart Downloads, where our app automatically downloads a premium user’s favourite tracks based on their listening history.

We have also invested significantly in hiring advanced AI & Machine Learning engineers who have designed autonomous recommendation engines that help us offer the most personalised music experience. Currently, more than 25% of the overall consumption is purely based on AI-ML led recommendations and we are hoping to double this in the coming year.

What about the Internet of Things? Devices like Alexa and Google Home are on the rise and IoTs are expected to mushroom.

Internet of Things will make music streaming even more seamless, and will open new avenues for audio device manufacturers to partner with streaming players to offer the most intuitive digital music experience. We are integrated in both Amazon Alexa and Google Home and look forward to be part of futuristic experiences being built on IoT platforms.