Shulou(Shulou.com)06/01 Report--

In late autumn, SequoiaDB Giant Sequoia Database brought you "a fire" in late autumn.

SequoiaDB v3.4 has been officially released!

Performance of distributed transaction scenarios is greatly improved

SequoiaDB Giant Sequoia Database version 3.4 has been officially released, and the most important feature of v3.4 is the performance improvement in distributed trading scenarios.

Compared with the previous version, SequoiaDB v3.4 improves read and write performance by 30%, update performance by 1-1.5 times, and query performance by more than 1.5 times compared with v3.2 in distributed transaction scenarios.

Performance comparison between new and old versions

For distributed trading scenarios, the technical improvements of version 3.4 mainly include the following:

Improved 2PC Algorithm

Distributed transaction intelligent arbitration algorithm. An intelligent arbitration algorithm is added to the 2PC submission of distributed transactions, focusing on solving the abnormal state of "In-doubt Transaction" in 2PC algorithm, realizing the intelligent arbitration of database for multi-partition transactions in extreme scenarios, and ensuring the strong consistency of distributed transactions.

Latch-less Memory Model

Implement multi-level memory pool and unlocked memory model. Database cluster pooled memory resources, memory pool multi-level management, session access achieved 99.99% memory access hit rate, achieved unlocked memory access in high concurrency OLTP scenarios, and system CPU utilization increased by 10%. SequoiaDB v3.4 provides both online memory monitoring and offline memory analysis capabilities to automatically generate memory analysis reports.

Improved Raft Algorithm

Break through the limit of Raft algorithm and realize full concurrency synchronization. SequoiaDB v3.4 introduces conflict arbitration mechanism, which breaks through the dilemma of serial synchronization when there is a unique key constraint in Raft algorithm, realizes full record-level concurrent synchronization between replicas, and greatly improves the efficiency of replica synchronization.

Improved Full-text Search Algorithm

Performance optimization of full-text indexing. Optimize the full-text index connection model, reduce the connection time and memory utilization, hit the full-text index of the count operation, directly through the index calculation results, greatly improve the count data reading performance.

At present, for financial transaction scenarios, giant sequoia database has been applied on a large scale in financial customers' core transactions, core downward movement, relational database replacement and other scenarios, including credit cards, online banking, loans, etc., in the application of financial transaction scenarios, leading the new generation of distributed databases in the industry.

SequoiaDB v3.4 feature improvement

This new version comprehensively improves the function and performance of financial-level transaction scenarios, and makes in-depth optimization in terms of distributed transactions, data consistency, concurrent CURD performance and SQL compatibility. In addition, in order to meet the strict technical requirements of financial-level trading scenarios for stability, SequoiaDB also upgraded the chaos testing framework, and the stability of the cluster has been greatly improved.

Other major updates for SequoiaDB v3.4 are as follows:

Storage engine

Transaction Auto-commit pushes down optimization, simplifies two-phase commit to one-phase commit, and improves transaction performance.

Transaction consistency confirmation mechanism

Implement multi-level memory pool and unlocked memory model

Full concurrent synchronization to improve the performance of replica data synchronization

Provide incremental data archiving and synchronization capabilities

By turning on the full mode and time mode of the log, the incremental data can be extracted, converted and archived by day or specified time, and the incremental data can be imported into other ODS systems.

Full-text indexing supports array types

Full-text indexing supports $or and $not operations

Performance of full-text indexing has been greatly improved

The access plan adds automatic expiration cleanup and enables parameterized caching of $in operations

Duplicate key substitution is supported for insert data

The index supports not null constraints

Optimize transaction monitoring performance, realize unlocked transaction monitoring mechanism, and reduce the performance impact of transaction monitoring management on external services.

SQL engine

Optimize high availability to achieve horizontal expansion of SQL engine

Operator pushes down the storage node to calculate accurately to improve the utilization of network bandwidth

Transaction Auto-commit pushes down the storage engine to simplify transaction two-phase commit into one-phase commit and improve transaction performance.

Support NO TRANSACTION mode to improve the performance of initialization data scenarios

Optimize DDL operations, including rename table,modify field,add primary key, index, etc.

Fully compatible with MariaDB syntax

Large object engine

Provide S3 compatible object storage interface

Large object storage supports automatic partitioning according to time order, improves the ability to access and manage large objects, and can quickly archive and clean up according to time.

Large object filtering supports filtering conditions and precise matching

Ease of use

Support the re-election ability of designated nodes

Provide SQL syntax to query the current status and monitoring information of the database

Provide performance monitoring and slow query analysis capabilities

The ease of use has been further improved, and the giant sequoia tool matrix has been officially launched.

With the release of SequoiaDB v3.4, Giant Sequoia Database has also been launched recently, a new generation of Giant Sequoia Database tool Matrix, and plans to release SequoiaPerf performance diagnosis tools in the near future.

SequoiaPerf performance tool will be released soon

Rich operation and maintenance management tools are the most intuitive and friendly experience for users to actually use the database. Giant sequoia database will continue to innovate and listen to the voices of community users. Make the best of user-oriented development, operation and maintenance, performance tuning tools and best practices.

At present, the number of large bank customers of Giant Sequoia Database has exceeded 50, and the application scenarios include core transactions, data center, content management and real-time data services. In the future, we will also maintain self-research and innovation and take the lead in distributed database technology and a variety of industry applications.

Go to the download center of Giant Sequoia database website to experience SequoiaDB V3.4.

Http://download.sequoiadb.com/cn/

To learn more, please click "here"

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.