Richard Ney talks ScyllaDB versus DynamoDB and Cassandra.
My name is Richard Ney. I am a principal with Lookout. I work on the team that deals with all of our data ingestion and query for all downstream services.
Because we deal with a large amount of data, we require databases, primarily a name-value keystore, to hold all of our information. In the past, we've used AWS services such as DynamoDB or S3 with Athena, but we wanted something that would scale and allow a reasonable amount of maintenance costs.
In the past, many people on my team have experience with the Cassandra database. But we didn't like Cassandra because the amount of maintenance requirements was very high, as well as the number of servers required.
We initially went with Dynamo DB, but as we scaled up in volume, we found that the per month costs of DynamoDB grows in a very linear manner and to the point where the cost would overwhelm our operating budget. So we started looking again at other databases.
I encountered ScyllaDB at the AWS reinvent conference showcase for two years that I attended, so I was aware of another database that provided all the functionality of Cassandra but was much lower maintenance. So we started looking at ScyllaDB, and we were very pleased with the operational characteristics. We were able to meet the requirements with a far fewer number of servers than we were expecting, and the resiliency and other maintenance issues that we encountered in the past with Cassandra were not there.
So we executed a large scale proof of concept. And we were able to make a decision that going forward direction, we'll be using ScyllaDB , as we love the ability to have all the flexibility of a Cassandra database with the performance and cost profile that meets our requirements.