Helping Financial Institutions Know Their Customers
Mumbai-based Karza Technologies is building a comprehensive digital due-diligence platform, helping financial institutions screen counterparties through more than 700 publicly available government information sources. Karza Technologies’ offerings help accelerate on-boarding, risk assessment, underwriting and due diligence, and monitoring processes.
We have lowered our IT costs because Lambda allows developers to write code for the infrastructure with no static allocation of servers.”Alok Kumar, Chief Technology Officer, Karza Technologies
A Switch to Serverless Computing Is Stifled
Karza Technologies launched its service with a cloud service provider (CSP) in 2015. After the company saw Amazon Web Services (AWS) introduce its AWS Lambda (Lambda) serverless compute service, it waited for its existing CSP to unveil a similar offering. Alok Kumar, chief technology officer of Karza Technologies, says, “We thought Lambda was such an awesome service because it allowed AWS customers to automatically and precisely scale applications just by uploading the code, reducing their IT costs.”
Even though the CSP used by Karza Technologies launched a serverless solution, it couldn’t offer the same level of support for the Python programming language as Lambda. Alok comments, “The support for Python in our CSP’s serverless solution wasn’t as mature as AWS’s.”
The Serverless Transition Gets Underway
Karza Technologies met with AWS solution architects to support a migration from the existing CSP to AWS. The business had IT personnel with experience in AWS spend a few weeks refreshing their knowledge of core AWS services and gleaned a quick understanding of the AWS Serverless Platform with special attention on Lambda, Amazon API Gateway and Amazon Elastic Container Service (Amazon ECS), with the objective of building highly scalable and secure applications. Alok says, “We designed our AWS infrastructure and then met with the AWS solution architects to rubber-stamp our plans before migrating our data to AWS. It took just a couple of months.”
Instant Scaling Secures High Performance
Karza Technologies’ MongoDB high-availability cluster, which stores data collated from various publicly available government databases, now runs on Amazon Elastic Compute Cloud (Amazon EC2) instances. Collated data is standardized using Karza Technologies’ in-house extract, transform, load (ETL) framework, which leverages Amazon ECS to orchestrate and auto-scale Karza Technologies’ data consumers. Standardized data is further ingested into a Hadoop-based Spark cluster, also running on Amazon EC2 instances. Here, the data is processed before ingesting into real-time databases such as Amazon Elasticsearch, Cassandra and JanusGraph.
The data is made available to Karza Technologies’ customers via the Amazon API Gateway and Lambda. In addition, Alok says that Amazon ECS is an important service because of the unpredictability of the quantities and speeds of the data coming in from the data sources. “Amazon ECS dynamically scales our virtual machines to cope with spikes in data traffic, so we don’t lose any performance on the front end.”
Reduced IT Costs, Faster Development Times
Karza Technologies has reduced its IT overhead by migrating to AWS. Alok says, “We have lowered our IT costs because Lambda allows developers to write code for the infrastructure with no static allocation of servers. It also allows us to focus purely on our business logic. Thanks to Amazon API Gateway, we created an API using Lambda with built-in security standards in one week—significantly faster and cheaper than we could on our previous CSP.”
Code Focus Drives Innovation
Karza Technologies is enabling greater innovation by using Lambda, Amazon Simple Queue Service (Amazon SQS) and Amazon Simple Notification Service (Amazon SNS). “We experiment more with ideas and try out new things because we’re using managed services like Lambda, Amazon SQS, and Amazon SNS and don’t need to invest time and money in provisioning and management,” says Alok. “To this end, we have implemented a distributed graph-based search engine that taps into hundreds of data points, all without managing or provisioning a single server.”