ScyllaDB

Hello everyone, I am searching for the appropriate solution for a financial use case. I want to handle user balancing in a distributed way. I did a little research and have found YugabyteDB as a solution that offers distributed ACID. Although, I have also found that ScyllaDB seems great in terms of TPS. So I wonder which solutions fit better, or maybe there is another great option I am not considering.

Please guys, what do you think?

At Pushnami we were using Redis for quite a long time, and still do for disposable caches. However as we continued to grow, we started to feel limitations of keeping data sets in memory, and only being able to use 1 CPU per server. We started to look around. Our requirements were pretty clear:

• Very fast reads and writes (sub ms)
• Ability to hold Terabytes of data
• Ability to horizontally scale
• Strong support for TTL and data eviction After looking at several alternatives we decided on @{Scylla}|tool:6895| , which had familiar Cassandra interface but claimed to be much faster. 2 years later - we run a cluster of 6 nodes that holds over 20 TB of data, and performs 150,000 operations per second. I would say pretty impressive! It was also easy to get started on AWS, as ScyllaDB provides prebuilt AMI with optimal configurations baked in.

ScyllaDB provides all of the goodies of Apache Cassandra, including HA, multiDC, replication sharding and so forth. The implementation is in C++ and the internal design is better and thus it achieves 10X the throughput, low 99% latency and more scylladb