Improving performance and lowering document database costs with AiSensy

AiSensy has experienced rapid business growth since it first launched its WhatsApp marketing platform in 2021. To process billions of messages, the company set up 12 clusters on MongoDB Atlas daily. 

As its data footprint grew to 120 TB, AiSensy began to search for solutions to reduce database costs and increase its storage limits. The company also experienced query performance issues on MongoDB, which impacted the stability of its platform and required ongoing maintenance. “We didn’t want our customers to lose trust in us,” says Gautam Rajesh Shelley, founder and CEO of AiSensy. “We experienced downtime at one point or another, so all of us were motivated to find a better solution.”

To be able to scale costs effectively, AiSensy decided to migrate to Amazon DocumentDB. Doing so would increase its platform’s scalability tenfold and help simplify the company’s technology stack on AWS. “Choosing Amazon DocumentDB was a no-brainer,” says Romit Singh Bhau, engineering manager at AiSensy. “If we migrated to some other document database, we’d have to rewrite a lot of our queries and application code, and that would’ve been a massive effort for our engineering teams.”

From the outset, AiSensy’s database team worked to maintain business continuity during the migration. Because the company was hitting performance and storage limits on MongoDB, AiSensy wanted to migrate as quickly as possible. 

To accelerate the migration, AiSensy used AWS Database Migration Service (AWS DMS), trusted by AWS customers to securely migrate more than 1.5 million databases. AiSensy worked closely alongside the AWS team to plan its migration. “We did a thorough evaluation of which clusters we needed to migrate first,” says Bhau. “We tracked our traffic patterns so that we could identify the ideal window to cutover.” 

AiSensy migrated the data from its clusters one by one. In between migration waves, the company performed load tests and data parity checks. To facilitate those efforts, AiSensy uses Amazon CloudWatch to observe and monitor resources and applications on AWS, on premises, and in other clouds. “All our logs are centralized in Amazon CloudWatch and can be accessed with a single query. It makes it very simple for our team to identify and troubleshoot issues,” says Bhau.

AiSensy uses Amazon DocumentDB change streams to automatically run event-driven functions using AWS Lambda, which empowers the company to run code without thinking about servers or clusters. When data is modified in Amazon DocumentDB, the change stream emits an event that cues an AWS Lambda function, which processes data asynchronously.

AiSensy completed the database migration in 3 weeks—with zero data loss or downtime. “Our migration went smoothly because we planned well in advance, and because we had consistent support from the AWS team,” says Bhau. “When we faced challenges, they were addressed proactively and effectively.”

The migration helped AiSensy realize a 20 percent savings on document database costs. The company also experiences better performance and can handle database queries 95 percent faster than before. “Since we migrated to Amazon DocumentDB, our customers can load their dashboards, track key analytics, retarget customers, and make critical business decisions much more quickly,” says Shelley. 

Adopting fully managed services has also fueled developer productivity at AiSensy. By architecting event-driven data pipelines on AWS, AiSensy’s engineering teams no longer need to perform daily cluster maintenance to effectively scale to customer demand. 

With these productivity gains, AiSensy has redirected its engineering resources toward new product development intitiatives, including building generative and agentic artificial intelligence (AI) features. For its generative AI use cases, AiSensy plans to deploy foundation models using Amazon Bedrock, which fast-tracks generative AI applications and agents from prototype to production. “One of the biggest benefits of the migration is that we can seamlessly integrate our database with other AWS services,” says Bhau. “Now our teams can set up internal data pipelines and build new product features for our customers faster than before.”