Home > News

SequoiaDB 3.0 Released, Leading NewSQL Database in China

Apr. 29, 2018

Recently, the leading NewSQL database SequoiaDB has released 3.0 version. SequoiaDB 3.0 has been in Beta from the late 2017, and has now been tested and applied in kernel online business systems of enterprise customers.

SequoiaDB 3.0 leads the new generation database technology innovation in industry.

SequoiaDB 3.0 Main Features

SequoiaDB is a financial-level distributed multi-model database that provide distributed NewSQL, distributed file system and object storage, and high-performance NoSQL storage modes, corresponding to distributed online transactions, unstructured data and content management , as well as massive data management and high performance access scenarios.

According to Gartner's database report, multi-model is the next major trend in the next decade of new-generation databases.

From high-performance distributed NoSQL database of SequoiaDB 1.0, to 2.0 unified distributed object storage of SequoiaDB 2.0, and fully transactional support and compatibility with MySQL of SequoiaDB 3.0, SequoiaDB has been continuously innovative and evolving.

1. MySQL Fully Compatible

SequoiaDB 3.0 is now 100% MySQL protocol-level compatibility:

• Full Compatibility: Full support for MySQL protocol and syntax, users can directly use MySQL client or any management, development and monitoring tools of MySQL to operate and access the database;

• MySQL Syntax: SequoiaDB 3.0 is 100% MySQL compliant with the MySQL's native parser. Supporting basic CRUD operations, multi-table associations, cross-node transaction operations, create views, stored procedures, indexing, and access plans etc.

• Smooth Migration: For any existing application, SequoiaDB 3.0 offers full MySQL compatibility and smooth migration without application code adjustments;

• Scaling: With the SequoiaDB native distributed storage engine, database is compatible with MySQL without the need for "sub-database partition tables". The distributed storage engine provides native scaling capability, which can increase the storage space and performance for more than 100 times.

• Multi-Dimensional Partition: With the “Storage-SQL” architecture, users can access database with multi-dimensional partitioning of the table and improve application flexibility.

SequoiaDB 3.0 uses a "Storage-SQL" architecture, which SQL layer and storage engine layer are independent of each other, similar architectures also appear on many new-generation distributed databases such as AWS's Aurora and Aliyun PolarDB.

SequoiaDB 3.0 uses the native SQL parser of the MySQL database, natively supports the MySQL protocol and is 100% syntactically compliant. In this architecture, the MySQL protocol parsing layer serves as the SQL parsing and distribution role and directly faces the application program.

Each MySQL service access node is a MySQL process that independently supports read and write operations. The data storage and management layer are entirely implemented by the database engine of the SequoiaDB.

SequoiaDB storage engine replaces the MySQL’s InnoDB engine, and natively support MySQL's syntax and functions, this provides the ability to expand the flexibility of the database storage layer.

2. Distributed OLTP Support

The MySQL compatibility of SequoiaDB 3.0 is mainly on its SQL syntax side, and many other data management mechanisms for distributed OLTP are implemented in SequoiaDB's distributed database engine.

SequoiaDB 3.0 brings some important improvements to distributed OLTP services in the storage engines:

ACID: ACID is the basis of a transactional database. SequoiaDB 3.0 has fully supported ACID and 100% support atomicity, consistency, isolation, and durability (ACID).

Cross-table/node Transactions: In a distributed database, atomic operations between multiple nodes need to be implemented in some special ways. SequoiaDB 3.0 uses two-phase commit to support cross-table or cross-node transaction capabilities.

Isolation: Support ‘read-committed’ isolation level;

Lock Mechanism: SequoiaDB kernel optimizes the management of record locks, completely avoiding the problem of lock congestion in a large number of concurrent transactions.

Cost-Based Optimization of Cost-Based Optimization (CBO): Implement statistical sampling of data and indexes within a set, establish a multi-dimensional, multi-level data model, and optimize external queries by "rewriting," "rule optimization," "parameterization," and "predicate degradation." Access performance, which is also the current optimizer optimization method for enterprise-class databases;

Compression: For table-level data compression, two compression methods are provided, with the highest compression ratio exceeding 60%, which greatly improves performance and throughput.

Security: The distributed architecture supports high-availability and off-site disaster recovery mechanisms. It also provides read/write separation mode while providing a main multi-standby storage. SequoiaDB 3.0 natively supports remote disaster recovery strategies such as the three centers in the two places to ensure the transaction data is safe and reliable.

Object Storage and File System Access

SequoiaDB 3.0 provides a standard Posix file system interface based on the object storage API. It can natively access any operating system that supports the Posix protocol standard, and users can migrate from NAS to SequoiaDB without any modification of the application.

Based on the use of the Posix file system, SequoiaDB 3.0 avoids the performance bottleneck caused by traditional file systems when storing large numbers of files. At the same time, with SequoiaDB's distributed architecture, its object storage and file system features provide scalable storage and concurrent throughput capabilities,which doesn’t need to change any of the applications.

In this version, the 'offset-lock' mechanism was also introduced. When concurrently operating the same file, each concurrent will lock only the contents of its offset, so that the correctness of the contents of the file under concurrent conditions can be guaranteed. Also, this will increased the concurrency of external access significantly.

Full-Text Search

SequoiaDB 3.0 provides full-text search capabilities. By creating a full-text index for a specific field, users can perform real-time fuzzy query on the content in the string, achieving the same full-text search capabilities as normal queries.

In the index type, a new 'full-text index' type has been added, and the user can easily create it by specifying the type of 'full-text index' when creating an index on the collection.

After the full-text index is created, the contents of the index definition are automatically synchronized to the full-text indexing engine, after which new data changes are quickly synchronized to the full-text indexing engine. SequoiaDB 3.0 supports asynchronous full-text indexing and does not have any impact on import performance when importing high-throughput data.

Disaster Recovery

SequoiaDB natively supports high availability at the database kernel level and cross-data center disaster recovery capabilities. It does not require the use of third-party tools to protect data with multiple copies, and fully meets financial-grade requirements:

• Highly Available: RPO, RTO is 0, and rapid response to seamless handover;

• Off-site Disaster Recovery: Disaster recovery and backup in different places to ensure data security. The distance between centers is more than 1000km. Satisfy the regulatory requirements of "three centers in two places";

• Active-Active: Real-time synchronization of dual-center data in the same city to ensure data consistency. Dual-center data can be read and written at the same time, greatly improving read and write efficiency. The central switching RPO is 0 and RTO is less than 1 minute.

• Easier DR Management: Unified management of disaster recovery centers in the system cluster simplifies maintenance costs and helps users get started faster.

Performance in OLTP Use Cases

SequoiaDB 3.0 has significant improvement in OLTP use cases. And the Sysbench benchmark results outperform the former versions. The benchmark test focus mainly on Insert, Query and Transactional use cases.

1)Insert:Single Insert

Running 3 96-concurrent Sysbench processes of Insert for 30 minutes.

The result shows that the average respond time is 5.28ms, TPS is 54,513.58 and 100% success rate.

2)Select:Indexed query, select a specific record in database.

Running 3 96-concurrent Sysbench processes of Select for 30 minutes.

The result shows that the average respond time is 1.42ms, TPS is 202,886 and 100% success rate.

3)OLTP Transactional:

Each OLTP transaction includes 10 precise-indexed queries, 1 indexed range query, 1 indexed range query summary, 1 indexed range query sort, 1 indexed range query deduplication, 1 precise match update index field, 1 exact match update non-index Field, 1 exact match delete, and 1 single record insert (Primary key is the value of the delete record field)

Running 3 96-concurrent Sysbench processes of OLTP transaction for 30 minutes.

The result shows that the average respond time is 68.74ms, TPS is 4,198.29 and 100% success rate.

Summary

SequoiaDB 3.0 covers all types of data structure and all business. Especially the strengthen in OLTP business, makes SequoiaDB a fully transactional multi-model distributed database.


About Us

SequoiaDB is a financial-level distributed database vendor and is the first Chinese database listed in Gartner’s Magic Quadrant OPDBMS report. SequoiaDB has recently released version 3.0.
SequoiaDB is now penetrating the vertical sector Financial Industry quickly and had more than 50 banking clients and hundreds of enterprise customers in industries including government, telecommunication, Internet and IoT.

Beijing:
Tower R, No.8 North Star East Road, Chaoyang District, Beijing,China
Guangzhou:
Tower A, No.22 Qinglan Street, Panyu District, Guangzhou,China
Shenzhen:
Tsing Hua Tech Park, Nanshan District, Shenzhen,China
Tel:400-8038-339
E-mail:contact@sequoiadb.com