VoltDB automatically partitions your data across your cluster, keeping multiple copies of all data to ensure high-availability and data safety.
- VoltDB scales partitionable transactions linearly with cluster size, up to millions per second.
- Unlike many clustered systems, VoltDB also supports full cross-partition, multi-round transactions.
- VoltDB can run up to tens of thousands of cross-partition read transactions per second and thousands of cross-partition write transactions.
- VoltDB is robust to node failures and network partitions and is designed to run at full speed with failed or rejoining nodes.
VoltDB was designed from day one as a shared-nothing, cluster-native, relational database.
- All nodes are the same; no leader or coordinator nodes.
- Send work to any node and it will be routed appropriately.
- No difference in functionality between single-node and clustered operation.
- One system to manage with unified monitoring and statistics.
- Read more about streamlined operations and TCO benefits.
VoltDB is focused on scalability and efficiency.
- Hundreds of thousands of ACID transactions per node.
- Millions of TPS on as few as three nodes.
- Each transaction can contain many SQL statements; run millions of write-heavy SQL statements per second on a single node.
- Read more about hardware efficiency.
As a result:
- Customers are primarily sizing clusters for data size and for desired redundancy levels.
- The most common cluster sizes are 3 nodes and 5 nodes, but many larger customers run 7 or 9 nodes per cluster.
- Our largest production cluster is about 60 nodes and we test clusters up to 100 nodes. We leave 1000 node headaches to less efficient systems.
- VoltDB does support multi-cluster replication, and many of our users have several small cluster working together across the world.