Recently, Tianyi Cloud HBlock launched a &#34;multi-active&#34; block storage service with dual support for multiple copies and erasure coding in its released version, achieving cross-region consistency and real-time data storage at the underlying level to ensure high availability of services. This capability has been deployed and operated in production-level resource pools in the Banan District, High-tech Zone, and Beibei District of Chongqing.
With the advancement of digitalization, &#34;no data loss, uninterrupted business&#34; has become a basic need for enterprises. Software vendors have provided various solutions for disaster recovery and backup, but still face issues such as high failure frequency and long recovery times.
Traditional block storage disaster recovery solutions are often limited to single cabinets, single data centers, or single availability zones. To achieve cross-availability zone disaster recovery, a complete set of resources and data must be built in a remote location, followed by data synchronization through software. This method is limited in real-time performance and carries the risk of single points of failure, requiring manual intervention for service recovery when issues occur. To ensure data security, multiple copies of redundant data must be stored across multiple availability zones, which increases costs; uncontrollable interruption durations severely affect customer experience and make it difficult to meet current enterprise requirements.
HBlock achieves strong consistency assurance across availability zones at the storage layer through protocol stack optimization and distributed algorithm innovation, constructing a high-performance block storage service with cross-availability zone RPO=0 and RTO in seconds. In the event of a failure in any single availability zone, the system can still achieve over 99.9% service availability (SLA). Through load balancing and multi-active architecture, business traffic can quickly switch to healthy available zones, making end users unaware of failures in any single availability zone.
According to the official introduction, HBlock has specifically achieved breakthroughs in three core capabilities:
<ol>
<li>Unified storage foundation, balancing reliability and efficiency</li>
</ol>
A unified metadata layer with real-time synchronization across regions; streamlined configuration, elastic expansion, and on-demand scaling; the storage pool can flexibly correspond to failure domains and business availability zones, employing erasure coding (EC) technology and real-time synchronization mechanisms to achieve strong consistency of data across multiple availability zones.
<img src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null"/> https://static001.geekbang.org/infoq/74/7481b5ed1a0c510fa6688204054be6d0.png<img src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null"/>
HBlock's multi-AZ/failure domain synchronization mechanism based on erasure coding (EC)
Business writes are synchronized across multiple availability zones, allowing for automatic seamless switching in the event of a failure in any region, ensuring continuous availability; replica folding (storing multiple copies in the same domain) enhances the fault tolerance of small-scale clusters while maintaining the disk utilization rate.
HBlock uses EC to stripe data across multiple availability zones, achieving a higher disk utilization rate compared to traditional three-copy synchronous or asynchronous replication solutions, and realizes RPO=0; it supports volume-level redundancy configuration strategies (such as three copies or EC N+M, N+M≤128) and various write modes (write-back, bypass, transparent write, SSD cache), allowing for customizable replica counts and shard sizes, enhancing performance and storage resource utilization
(2) Intelligent Self-Healing Engine, Seamless Switching Throughout
When a certain available zone fails, the self-healing engine instantly perceives and automatically pre-allocates channels, switching I/O to a healthy available zone, and triggering cross-AZ data reconstruction without manual intervention, ensuring stable and continuous service; the distributed multi-control architecture achieves automatic migration of volume connections and resets primary and secondary priorities, maintaining co-domain deployment of computing and storage, optimizing performance loss; after the failure area is restored, it can automatically switch connections back to the original available zone.
<img src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null"/> https://static001.geekbang.org/infoq/65/655ce3aa184c6b9c78ecee2d7da30632.png<img src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null"/>
<img src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null"/> https://static001.geekbang.org/infoq/ee/eee82ea52fa9a6884c41a82c8eabd318.png<img src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://www.infoq.cn/news/null"/>
HBlock's Self-Healing Engine Mechanism (Taking 3AZ as an Example)
(3) Intelligent Perception Scheduling, Continuous Service in Chaotic Environments, No Data Loss
HBlock is specially designed and validated for chaotic environments such as weak networks, weak computing, and weak disks. It does not rely on clock services and quickly triggers second-level fault detection and data reconstruction based on strong consistency arbitration; it dynamically collects multi-dimensional indicators such as storage load and network performance, and adjusts replica distribution in real-time through intelligent scheduling algorithms, eliminating slow disks and avoiding bottleneck nodes, ensuring data consistency and high availability of services.
HBlock is the industry's first lightweight storage cluster controller independently developed by Tianyi Cloud, capable of activating heterogeneous asymmetric storage resources with a low threshold, supporting self-service download of minute-level delivery software-defined storage products from the official website. Its flexible characteristics support deployment in various environments such as physical machines, virtual machines, and containers, allowing users to experience HBlock's flexible networking and rapid delivery of software-defined storage capabilities through tenant machine VPC.
This time, through technologies such as remote redundancy, self-healing engines, and intelligent scheduling, data is automatically distributed to multiple remote available zones, achieving real-time remote multi-active collaborative capabilities. Combined with existing features such as storage gateways, snapshots, and clones, it forms a multi-dimensional service guarantee of &#34;local high availability + remote multi-active + multi-version backup + hybrid cloud disaster recovery,&#34; providing new solutions for industries such as finance, government, and transportation

Tianyi Cloud HBlock launched the "Multi-Active in Different Locations" block storage service, supporting multiple copies and erasure coding to ensure high availability of services. This service has achieved production-level deployment in three resource pools in Chongqing, addressing the high failure frequency and long recovery time issues of traditional disaster recovery solutions. HBlock achieves strong consistency across availability zones by optimizing storage protocols and distributed algorithms, providing high-performance storage services with RPO=0 and RTO in seconds, ensuring over 99.9% service availability

Tianyi Cloud HBlock officially launches "Multi-Active" block storage