rapidgugl.blogg.se - Dynamodb table example

throughput can scale almost infinitely without throttling as long you don’t exceed 2X your all-time peak traffic inside of a 30 min span.

writes and reads to Dynamo are highly efficient - averaging single millisecond latencies - making it ideal as a high read and write datastore.

DynamoDB tables with OnDemand capacity offered the perfect intersection of these requirements. We also wanted this solution to be managed externally, something that we could spin up and not worry about, allowing us to reduce our operational overhead. Scalability: We wanted a solution that could scale quickly as we grew.

In the end, we chose DynamoDB for a couple of key reasons: scalability, isolation, and flexibility. We explored a variety of different solutions from MongoDB Atlas, Apache Cassandra, and Couchbase. This model, which powers our fraud detection pipeline, was not relational and was growing rapidly due to the increasing number of transactions processed through Bolt. We had a variety of data models that naturally fit in a No-SQL paradigm, but our transaction risk data was a great first use case.

In late 2020, we set out to find a good No-SQL database to complement our Postgres DB. We chose DynamoDB over a variety of well established No-SQL technologies