At a glance
This global retailer utilizes a large set of microservice-based apps to manage the product catalog for their e-commerce website. These apps enable customers to shop anytime, anywhere, both online and in stores.
Their Cassandra database’s level of transactional consistency and the read and write performance was inadequate. As a result, customer and vendor use cases were complex to implement and maintain.
YugabyteDB’s ability to support consistent RDBMS-style transactions and a NoSQL programming model on the same datastore met their key requirement for high availability, performance, and resiliency.
Retailer cost-effectively supports product catalog with over 100 million items and 100,000+ queries per second
Focused on constant innovation, this global retail corporation operates a chain of stores as well as an ecommerce website that depends on a set of business-critical microservices to manage its expansive product catalog and key online shopping functions.
The retailer originally migrated from Oracle to Cassandra to help support exponential data growth and performance needs. However, they encountered problems with data consistency.
The lack of distributed transactions and RDBMS capabilities in Cassandra forced them to implement one-to-one and many-to-one bidirectional indexes in the application layer. This increased costs and delayed getting new services into market.
Key selection criteria for a new database included resilience, high availability, and performance that could cost-effectively scale with growth. The ability to deploy the database in any cloud in a geo-distributed topology across regions and countries was also important. Other considerations included support for multi-key ACID transactions and alternate key lookups.
Challenges
- Cassandra unable to guarantee data consistency in the product catalog, costing millions of dollars annually as the retailer was forced to use expensive 3rd party vendors to fulfill promised orders.
- Lack of an easy, multi-region deployment model left the retailer at risk during any type of regional failure, such as a catastrophic weather condition due to an ice storm
- Complex operations and database limits impacted their ability to cost-effectively grow and scale.
Key Database Requirements
The retailer selected distributed SQL database, YugabyteDB, because it supports:
- Very high-volume ACID-transactional consistency across multi-region deployments
- Cost-effective liner scalability that could keep up with growth
- High availability and performance along with strong resiliency
- Flexible datacenter topologies, with multi-cloud and multi-region support
- Cassandra-compatible API to accelerate the modernization of existing Cassandra-native applications
- Familiar tools and APIs to minimize learning curves and accelerate productivity
“We need a data architecture that can scale to billions of mappings for millions of objects and serve hundreds of thousands of requests. YugabyteDB is the answer.”—Sr. Technology Leader
Business Results
100M+
items supported in the product catalog
$10M
in lost revenue recovered
250K+
queries per second achieved
YugabyteDB Solution
RDBMS capabilities including multi-row ACID transactions and alternate key lookups
Cloud-agnostic, geo-distributed database with an open-source licensing model
Strongly consistent and resilient database that can be deployed across multi-datacenter topologies.