Automatic Zstd compression at the page level.
True in-memory tables: all data is permanently stored in memory, not just the write-ahead buffer.
Picodata’s LSM-tree: more compact metadata storage and efficient automatic garbage collection. REPAIR is no longer required.
A single data space: all nodes see the entire schema and all data.
The equivalent of zero-token nodes thanks to the use of a tiered architecture.
Efficient, consistent secondary indexes.
FSTEC certification at trust level 4.
Tested for industrial MES/IIoT workloads.
Horizontal scaling with built-in fault tolerance.
Strict consistency by default, with no trade-offs between speed and reliability.
Record-high transaction throughput: lightweight transactions (LWT) are faster than in Cassandra.
Optional in-memory storage: response times are tens of times lower than in Cassandra.
Eventual Consistency
Strong Consistency
Hybrid
(in-memory + LSM-tree)
Sirin is a next-generation distributed DBMS, fully compatible with Apache Cassandra and built on top of the Picodata DBMS. The product is intended for the industrial sector, government information systems, and infrastructure where reliability, consistency, and performance are especially important.
The Picodata DBMS is included in the unified register of Russian software and is developed in Rust, which guarantees security and high efficiency. Sirin inherits all the advantages of Picodata’s architecture and complements them with compatibility with Cassandra and CQL, making it an ideal choice for companies looking for a Cassandra equivalent without needing to change application code.
Sirin is fully compatible with CQL (database) and Apache Cassandra drivers, but eliminates the key limitations of the original solution. At its core is a strongly consistent (Strong Consistency) data store implemented with a shard-per-core architecture, an extreme approach to horizontal scalability in which each core of a multi-core processor is responsible for its own data segment. Lightweight transactions (LWT), implemented using Raft rather than the Paxos algorithm used in Cassandra, execute faster than in Cassandra. Like Cassandra, Sirin supports automatic Zstd compression, which reduces disk load and improves system responsiveness.
Sirin is a distributed database capable of storing data either in RAM or on disk, depending on the workload profile. This approach ensures instant access and stable performance even under peak loads.
At the core of Sirin is an enhanced LSM-tree architecture from Picodata that provides compact metadata storage and fast automatic cleanup without the need for manual REPAIR. If uncompromising data processing speed is required, the system uses true in-memory tables, where all data is permanently stored in memory, making operations virtually instantaneous.
The shard-per-core architecture distributes the load across CPU cores, allowing you to:
Sirin unifies all nodes into a single data space: each node sees the full schema and has access to all tables. This accelerates distributed query execution and increases cluster fault tolerance.
Sirin is a Cassandra equivalent designed for environments where data loss and downtime are unacceptable. The system implements an ACID approach: any changes are first written to a log on disk and replicated to multiple servers. In the event of a node failure, the data is automatically restored.
This level of fault tolerance makes Sirin ready for operation in industrial MES/IIoT systems and critical IT environments.
Picodata Sirin is certified by FSTEC of Russia (trust level 4), which allows the system to be used in government and corporate information environments with increased security requirements.
The solution is developed by Russian engineers, fully supported within the Russian Federation, and included in the register of domestic software. For companies using Apache Cassandra, migration of Cassandra applications is seamless: Sirin provides full compatibility at the interface and CQL query level.