Share Chat: Building a Scalable Data-Driven Social Network for Non-English Speakers Globally

India is a multilingual nation with 22 main official languages and numerous additional regional dialects that are spoken in rural regions. There are currently more than 227 million internet users in rural parts of India, and this number is anticipated to rise dramatically as internet connectivity becomes more widely available.

Indians in rural areas and people from all over the world can connect on ShareChat, which was conceptualized, planned, and built in India.

In 15 different Indian languages, more than 160 million active users per month share and view movies, pictures, GIFs, songs, and much more. In July 2020, ShareChat also debuted Moj, a short video platform with over 80 million monthly active users that has since established itself as the top short video platform in India.

ShareChat makes it easier to find content and people by employing a tailored content feed as the front page of its mobile app, knowing that many new internet users don’t know how to utilize search terms to find content. The business’ data science team use machine learning models to identify the language of the content and user involvement in order to surface the appropriate content to the appropriate user.

With a Few Clicks, Apps May Be Scaled to Handle Unforeseen Demand


Data is the driving force behind ShareChat’s application. ShareChat uses real-time data to power its social network platform, monitoring every user behavior within the app, from chat messages and the creation of new groups to what users enjoy and who they follow.

ShareChat requires its systems to process terabytes of data at extremely high levels of efficiency since its users publish more than a million posts per day. Because of Cloud Spanner’s global consistency and secondary indexing, ShareChat moved from a NoSQL database to it.

ShareChat reduced costs by 30% by moving 120 tables with 17 indexes from 220 database tables to Cloud Spanner, which supports transactional workloads. ShareChat can easily recover a copy of any documents they need, even if one area fails, thanks to Cloud Spanner’s ability to effortlessly duplicate data in several places in real-time.

In just a few days, the company’s traffic increased by 500 percent, and it was able to extend horizontally with no code changes. This occurred concurrently with the Moj app launch, and it was seamlessly expanded to a different region within a week. The app was able to handle the additional demand that was placed on it thanks to Cloud Spanner without any assistance.

Read More- How to make a Blast Furnace in Minecraft

With No User Downtime, “Big-Bang” Migration

In order to ensure a seamless migration, ShareChat put together a dream team with the appropriate resources and abilities. ShareChat’s engineers designed, tested, and carried out the migration procedure jointly with CloudCover, a partner with experience in large-scale cloud migrations.

ShareChat already had 220 tables and more than 70 terabytes of data when it was time for the conversion. Some of these tables have about 50 billion rows and take up 14 gigabytes of space.

Instead of a gradual deployment, ShareChat chose a big-bang migration. Due to the interdependencies of data, users may encounter latency spikes from out-of-sync data, which would impede its timely delivery, if services were transferred one at a time. For instance, users may receive messages after the fact, such as invitations to events that are two weeks old, if the data is out of sync. Such errors could make users discontinue using ShareChat.

Despite moving gigabytes of data, Venkatesh asserts that none of his clients should be impacted. “Google Cloud gave us a lot of confidence that this could be done and that it would work with us as a partner. With our past cloud partners, we didn’t experience this.

Since the platform must support more than one million queries per second, ShareChat ran a proof-of-concept cluster for four months to test database performance in a real-world setting. To conduct thorough performance tests and capacity analyses, it replicated all data from the legacy system to Google Cloud using an open source API gateway.

Read More- How to Fix DoorDash Promo Codes Not Working

Enhancing the Business Ads on Share Chat

The next stage of ShareChat’s development involves improving how it monetizes its service to provide value to customers as well as its investors. ShareChat is now available for free download. Ad-based income is now generated by the app to fund further app development. To develop regional language advertising, the business links brands with content producers.

Through the use of hyper-local content on ShareChat, partners hope to improve the brand experience of rural consumers. The Business and Strategy team uses BigQuery and Data Studio to visualize campaign performance information to assist marketers in maximizing marketing expenditures.

To visualize resource utilization recorded in GKE usage metering, a Google Kubernetes Engine feature, ShareChat additionally makes use of BigQuery and Data Studio. To track over-provisioned resources and cut waste, ShareChat may publish a cost breakdown for several departments by simply tagging the services in BigQuery.

Read More- Vtuber Maker & How-To: Get Your Own Vtuber Avatar

Automation for DevOps simplification


The average throughput for ShareChat is 80,000 requests per second (RPS), or around 7 billion RPS per day. Given the scale of its subscriber base, push alerts like the daily trending topic sent to users at 5 PM can cause an instantaneous increase in RPS of 130,000.

To avoid over-provisioning servers, Venkatesh explains, “we can pre-scale Google Kubernetes Engine for traffic spikes during scheduled holidays like Diwali when millions of Indians send greetings.” “Moving to a native Kubernetes environment substantially increases our capacity to embrace agile work practices, such as automated deployment and to save time writing scripts.”

ShareChat can attach ancillary activities like logging into the application without modifying any code thanks to special Kubernetes features like sidecar proxy.

Google Kubernetes handles Kubernetes upgrades by default, unlike the prior cloud provider. For the purposes of reducing security risks and utilizing new capabilities, clusters and nodes automatically upgrade to run the most recent Kubernetes version.

Comments are closed.