Skip to main content



The Overall Design Objectives of CnosDB 2.0

Design and develop a high performance, high compression ratio, high availability of distributed cloud native time series database to meet the following goals:

Time Series Database

  1. Scalability: Theoretically supports an unlimited number of time series, completely solving the problem of time series inflation, and supports horizontal/vertical scaling.
  2. Computing and storage separation: Compute nodes and storage nodes can be independently scaled up or down, with sub-second scalability.
  3. High-performance storage and low cost: Utilizes a high-performance I/O stack, supports tiered storage using cloud disks and object storage.
  4. Query engine supports vectorized queries.
  5. Supports multiple time series protocols for writing and querying, provides external components for data import.

Cloud Native

  1. Supports cloud native, leveraging the convenience brought by cloud infrastructure and integrating into the cloud native ecosystem.
  2. High availability: Sub-second fault recovery, supports multi-cloud, cross-region disaster recovery.
  3. Native support for multi-tenancy, pay-as-you-go.
  4. CDC (Change Data Capture): Logs can be subscribed to and distributed to other nodes.
  5. Provides more configurable options to meet the complex needs of public cloud users.
  6. Cloud-edge collaboration: Provides the ability to integrate edge devices with public clouds.
  7. Integrates with cloud-based OLAP/CloudAI data ecosystems.

The Design Objectives of CnosDB 2.4

CnosDB V version 2.4.0 will achieve: add timing functions, add lossy compression algorithms, support update and delete operations and have a master replication group, this version of CnosDB will be able to provide users with a higher performance, easier to operate, more comprehensive data processing capabilities of the open source timing database, described in detail below.

The Space-Time Function

Support space type and space-time function, space type will be supported in the standard WKT and WKB mode, supported space-time function including but not limited to position calculation, area calculation, moving speed calculation, etc.

Advanced Functions

Common functions (first, last, max, min), date conversion class, monitoring class (gauges calculation).

Lossy Compression

  • Deadband Compression: A data compression algorithm used to reduce the frequency of sensor data updates and reduce data transmission and storage costs.
  • Swinging Door Trending (SDT)  Algorithm: A real-time data stream processing algorithm that can be used to process dynamic data sets, maintaining the number of elements in the data set by constantly adjusting the size of the gate.

Support Schema change

Support Update、Delete operations.

Add Master-Slave replicaset

Implements exactly once semantics for processing stream data.