Gamoshi Improves Business Intelligence Using Amazon Kinesis, Amazon Redshift, and Upsolver

Digital advertising and marketing technology company Gamoshi needs to be efficient for its programmatic advertising model to succeed. Programmatic advertising—the automated buying and selling of online advertising—relies heavily on real-time access to large amounts of data, so agility and business intelligence can help a company differentiate itself from its competitors. Gamoshi was looking to optimize its offerings and provide its customers with the tools they need to achieve their business objectives. To meet these goals, the company chose to build new infrastructure and business offerings on Amazon Web Services (AWS), improving efficiency, improving business intelligence, and ultimately saving money for Gamoshi and its customers.

Founded in 2016 in Israel, Gamoshi provides customers, who range from small businesses to large enterprise companies, the ability to join an online private marketplace, an invitation-only location in the cloud where select buyers and sellers do business. Gamoshi supports billions of transactions daily with up to 400,000 real-time bidding queries per second, simultaneously helping advertisers to reach their target audiences and publishers to monetize their inventory. Because Gamoshi seeks to provide customers with full transparency, it is critical for the company to collect, store, and prepare data reliably in near real time while minimizing cost. Gamoshi needed a solution to drive business intelligence that would combine data analytics, data visualization, and a cloud-based infrastructure to help its customers make data-driven decisions.

Gamoshi had been running Kubernetes clusters on a third party’s cloud infrastructure, but its customers did not have access to all the data they wanted without having to pay extra. In July 2020, Gamoshi turned to AWS because of its reputation for providing cloud-based infrastructure and being available 24/7. The company started with basic services such as traffic routing and then began transferring data from its previous cloud provider to AWS. At first, Gamoshi maintained two parallel clusters to test that its system was working properly. Then, after rigorous testing, the company began a more comprehensive migration to AWS. “We got better performance than what we had using our previous cloud provider,” says Moshe Moses, Gamoshi’s cofounder and CEO. “Plus, I’d say we had more than a 30 percent cost reduction.”

So far, Gamoshi has migrated 90 percent of its infrastructure and data to Amazon Redshift, a data warehouse that makes it simple to gain new insights from data. Amazon Redshift provided Gamoshi with the performance it was looking for without losing any granularity in its data model. “The main thing I like about Amazon Redshift is how simple it is to change the data structure,” Moses says. “It’s a very simple and elegant solution to a big problem.”

Gamoshi uses Amazon Redshift to provide customers greater range when querying data. Under the previous system, a customer couldn’t query more than 3 months ahead without it becoming cost prohibitive. But by using Amazon Redshift, customers can now query more than a year ahead, obtaining a wider range of metrics rapidly and reliably without additional cost. Plus, Gamoshi can clean, merge, and insert its data into another table in 2–3 seconds, a process that used to take up to 60 seconds. Time to market for big data, which used to take 1–2 weeks, now takes only 1 day including tests. “For the users, it’s like magic,” Moses says. Also, Gamoshi was able to use Amazon Redshift without having to change its third-party application Upsolver, a fully managed data lake solution that is available on AWS Marketplace.

In addition to Amazon Redshift, Gamoshi uses Amazon Managed Service for Apache Flink, which provides a simple way to transform and analyze streaming data in real time with Apache Flink. Using Amazon Managed Service for Apache Flink, Gamoshi can separate data records belonging to a stream into multiple shards and derive insights from data cheaply, facilitating frequent use of the service. In addition to its affordability, the reliability of the service has proven impressive to Gamoshi. “Even if we have three times more traffic in a couple of days, I know it will handle it,” Moses says. “We don’t have any downtime, and it’s going to be completely transparent. We can be silent, efficient, simple to use, and much cheaper than I thought.” To further optimize performance and cost, Gamoshi uses AWS Graviton Processor, which delivers the best price performance for cloud workloads running in Amazon Elastic Compute Cloud (Amazon EC2), a web service that provides secure, resizable compute capacity in the cloud. After a short learning curve for developers, Gamoshi was able to reduce its regular staffing by 20–30 percent, saving even more.

Gamoshi also benefits from assistance from its AWS account manager and AWS solutions architect, who provide technical details and support for AWS solutions. “Their support is one of the main things that helped us migrate,” Moses says. “One of the good things about AWS is that it has so many solutions, and I know that I can pick up the phone and talk to someone at any time.”

Because Amazon EC2 instances are hosted in multiple locations worldwide, Gamoshi is looking to deploy in additional Availability Zones and explore AWS Global Cloud Infrastructure, a secure, extensive, and reliable cloud solution. Gamoshi also wants to further explore data analytics and machine learning on AWS to improve its business intelligence and power decision-making through the analysis of raw data. Additionally, for its customers, Gamoshi hopes to offer more predictive tools through an enhanced user interface. “We are the most efficient we’ve ever been,” Moses says. “It’s only the beginning. Using AWS, I don’t think we have any limitations to our growth.”