Release Note

Release Note: MatrixOne 1.1.0

Posted by MatrixOriginPublished on

We are excited to announce MatrixOne 1.1.0 release!

MatrixOne is a hyper-converged cloud-native database. It is designed to provide a cloud-native, high-performance, highly scalable, MySQL-compatible HTAP database. MatrixOne enables users to handle mixed workloads such as transactions, analytics, time-series, and streaming processing through a one-stop data processing solution.

What's New in v1.1.0?

Vector Data Type

These features enable users to quickly build AI applications, such as RAG applications based on large language models (LLMs). Unlike specialized vector databases, MatrixOne is a general database highly compatible with MySQL, enabling users to get started quickly without additional learning barriers. It also integrates structured and vector data processing for building AI applications.

  • Implemented vecf32 (float32) and vecf64 (float64) type.
  • Support for basic binary operators: +, -, *, /.
  • Support for comparison operators: =, !=, >, >=, <, <=.
  • Support for unary functions: sqrt, abs, cast.
  • Support for vector functions: summation, l1_norm, l2_norm, vector_dims, inner_product, cosine_similarity.
  • Support for aggregate function: count.

Time Series

  • Support for streaming loading with LOAD INTO INLINE, surpassing INSERT INTO in performance.
  • Support for time-series tables with timestamps as primary keys, and support for any dimension/metric columns.
  • Support sliding window for downsampling queries over different time periods.
  • Support for interpolation with various interpolation methods.

Kafka Connector(beta)

  • Support for creating dynamic and append-only table with CREATE DYNAMIC TABLE.
  • Support for configuring external data sources with CREATE SOURCE.
  • Support for Kafka topic integration with JSON or protobuf format.

User Defined Functions(beta)

  • Support for creating Python-based UDF

Other New Features

DDL Statements

  • Support for insert on duplicate key ignore.
  • Support for create or replace view.
  • Support for alter sequence.
  • Support for Key, hash partition pruning capabilities (beta).
  • Support for List/List column, Range/Range Columns partition storage capabilities (beta).

Indexes and Constraints

  • Full support of secondary indexes for dynamic query acceleration.

Built-in Functions

  • Added SAMPLE sampling function.
  • Added CONVERT_TZ time zone conversion function.
  • Added SHA2 encryption function.
  • Added Encode/Decode encoding and decoding functions.

Security

  • Support for managing path permissions for select into through creating Stage.

Tools

  • mo_dump tool (logical backup)
  • Starting from this iteration, modump is managed in a separate repo (https://github.com/matrixorigin/mo_dump).
  • Supports exporting DDLs separately.
  • Supports exporting multiple databases and tables.
  • mo_backup tool (physical backup)
  • Supports file systems and object storage as storage media for backup and restoration.
  • mo_ctl_standalone Tool
  • Supports automatic data backup.
  • Supports automatic log table data cleaning.
  • Supports converting data files from CSV format to insert or load data inline format.
  • Supports automatic docker image building.
  • Supports docker mode for standalone deployment.
  • mo_ctl_distributed Tool
  • Supports one-click installation and uninstallation of distributed clusters.
  • Supports start/stop, upgrade/rollback operations for matrixone clusters.
  • Supports installing matrixone clusters in different k8s clusters.
  • mo_operator tool
  • Supports configuring custom S3 certificates.
  • Supports backup and recovery, and management of backup data for matrixone clusters via API.
  • Supports setting optimized Go GC strategies automatically for matrixone clusters.
  • Supports enabling Python UDF for matrixone clusters.
  • Supports integration of matrixone on Kubernetes with Prometheus.

MySQL Compatibility

  • Remove hundreds of MySQL-incompatible reserved keywords.

Known Issues

  • Memory leak occasionally happens and may lead to an OOM error.
  • Occasional system hung under high concurrency workload.
  • Kafka connector works only in a standlone deployment.
  • Secondary Index doesnt apply for IN queries.

New Contributors

Full Changelog: v1.0.0…v1.1.0