Abstract: HUAWEI CLOUD DCS Redis 6.0 Community Edition brings cloud services with extreme performance, comprehensive functions, strong reliability, and cost-effectiveness, and is fully compatible with open source Redis. The client does not need to modify the code and can be used after activation, enabling enterprises There is no need to worry at all, you can enjoy the gold revenue that has increased the business response speed several times.
This article is shared from the HUAWEI CLOUD community " DCS Community Edition Redis6.0 Technology Secret ", author: jingle cat.
Since Redis entered the 6.0 version, new features and functional improvements have been changed every month, and the upgrade speed is simply amazing! And, for version 6.0, Antirez, the father of Redis, used several "EST" words on his blog when the RC1 version was released (2019-12-19):
This version provides many exciting new features and functional improvements, such as the new network protocol RESP3, the new cluster proxy, ACL, etc., among which the most concerned one should be "multithreading".
Huawei Cloud DCS also launched support for Redis 6.0 for the first time. After a lot of preliminary work preparations, Huawei Cloud DCS Community Edition Redis 6.0 was released in early August 2021, and is in public beta testing . At the same time, compared with the open source Redis6.x, the DCS Community Edition Redis6 is 1.5 to 3 times the performance of the open source version.
How does it do it? Let's start a chat for everyone.
DCS Community Edition Redis6.0 Product Performance
Not much to say, the picture above:
Performance comparison test
As shown in the figure, in the case of 400 client connections, with 2 threads, DCS write performance is 1.68 times that of open source, and read performance is 1.54 times that of open source. The latency is 39% and 35% faster than open source respectively; when 4 threads, DCS write The performance is 2.56 times that of open source, and the read performance is 2.22 times that of open source; the latency is 61% and 55% faster than open source respectively.
Performance improvement analysis
The judges can see that the performance of DCS Community Edition Redis 6.0 has been greatly improved, so how does it do it? Listen to the little brother slowly.
In the Redis solution, the I/O thread task only reads and parses the client request command through the Socket, but does not actually execute the command. All client commands need to be executed in the main thread at the end, so the use of multi-core The rate is not high, and every time the main thread must be busy polling after all the tasks are assigned, waiting for all I/O threads to complete tasks before continuing to execute other logic. Redis designed its multi-threaded network model in this way. I think the main reason is to maintain compatibility and to use multi-core to improve I/O performance. It should be a compromise choice.
HUAWEI CLOUD DCS Redis has achieved true multi-threaded optimization and improvement. In addition to multi-threaded network concurrency, it also optimized the multi-threaded event processing mechanism, which increased our resource utilization and performance gains by 2 to 3 times. In addition, vertical elastic scaling can also be more hierarchical and so on.
In-depth comparison with the open source version
The following table is a detailed comparison between the DCS community version Redis 6.0 and the open source version:
To sum up: HUAWEI CLOUD DCS Redis 6.0 Community Edition brings cloud services with extreme performance, comprehensive functions, strong reliability, and cost-effectiveness. It is also fully compatible with open source Redis. The client does not need to modify the code and can be used after activation, making the enterprise completely There is no need to worry about the benefits of gold that has increased the business response speed several times.
, DCS Redis6.0 community version is now on the public beta , I look forward to everyone's enthusiastic registration, the number is limited, first come first served. 【Small speaker】
- Redis author Antirez's blog: http://antirez.com