Abstract: During Huawei Fully Connected 2021, Huawei cloud GaussDB(for openGauss) officially launches a major new kernel feature, which is the in-city dual-cluster HA solution. It provides final-level HA services and supports the in-city dual-cluster switchover with RPO=0 and RTO<60s. Ensure the core business of enterprise users without interruption for 7X24 hours as much as possible.

During Huawei Fully Connected 2021, Huawei cloud GaussDB(for openGauss) officially launches a major new kernel feature, which is the in-city dual-cluster HA solution. It provides final-level HA services and supports the in-city dual-cluster switchover with RPO=0 and RTO<60s. Ensure the core business of enterprise users without interruption for 7X24 hours as much as possible.

GaussDB for openGauss (for openGauss) is a high availability technology in a new architecture and deployment mode. In existing HIGH availability (HA) solutions, a single cluster with multiple copies is deployed across AZs. Therefore, faults cannot be isolated from each other. As a result, faults of cluster management components or other regional faults may cause service unavailability in the entire cluster. For the traditional network-based log synchronization method, the increase of the geographical distance between the primary and standby nodes of the database will lead to a significant increase in transmission delay, which directly affects the performance of production services. At the same time, the financial and banking sectors have high requirements for data security, which requires maximum assurance of data security and service availability. Therefore, GaussDB provides a high availability solution with RPO=0 for the database in the same city and two clusters. That is, when the primary cluster is faulty, the secondary cluster can continue to provide services. When a natural or man-made disaster occurs, the secondary cluster protects data and rapidly recovers data without tolerating data loss.

This section describes the GaussDB in-city dual-cluster HA solution

The active and standby database clusters are deployed in two azs in the same city and support centralized and distributed architecture. The active cluster serves as the production center to provide production services, and the standby cluster serves as the hot spare center to provide read-only and failover functions. Redo logs in the primary cluster are synchronized to storage devices in the standby cluster using the storage layer data replication technology. The standby node reads Redo logs from storage devices in the standby cluster and plays back the Redo logs. Transactions on the primary node are committed only after the logs written by the primary database node are synchronized to the storage devices in the secondary cluster, thus ensuring the performance indicator of cluster switchover RPO=0. The storage device uses huawei’S OceanStor Dorado V6 all-flash storage system. It can remotely replicate data in parallel, and provides stable end-to-end performance with low latency and high throughput.

In addition, the solution has good scalability. In addition to the same-city dual-cluster HA solution, the solution can also form cross-region Dr With remote clusters. That is, a remote Dr Center is added to back up data of the same-city dual-cluster solution to form a two-site and three-center Dr Solution.

The two-cluster HA solution in the same city integrates the GaussDB database and shared storage devices. It takes advantage of the performance and reliability of shared storage devices to achieve a cross-cluster high availability architecture of RPO=0 and provides more enterprise-level functions for the GaussDB(for openGauss).

Schematic diagram of distributed architecture of dual cluster ha solution in the same city

Core advantages of GaussDB in-city dual-cluster HA solution

  • Financial high availability: Supports switchover between two clusters in the same city with RPO=0 and RTO<60 seconds, ensuring service security and reliability. When the primary cluster is faulty, the secondary cluster can quickly complete data switchover without damage and continue to provide production services.

  • High performance: First, the performance of physical log synchronization is 10 times higher than that of logical log synchronization. Second, through Dorado storage hardware to achieve fast synchronization of logs between clusters, using Dorado inherent network protocol (dense wave division), reduce network delay more than double, while using Dorado storage cache capacity, log write instant persistence, reduce transaction submission delay.

  • High reliability: data security to achieve double insurance, on the one hand, the database kernel copy to ensure automatic failover and recovery, no interruption of business; On the other hand, the storage kernel ensures disk sub-health, fault tolerance, hardware self-healing and other capabilities.

  • Advanced architecture: The internal computing and storage of the database are separated, and storage management is placed in the lower-layer shared storage, which solves the delay caused by data synchronization and increases the horizontal scalability of computing capability.

  • Cluster isolation: Database clusters are decoupled and fault domains are isolated to avoid global network failures and cluster management component failures. Version isolation between clusters, avoid Bug pollution, can quickly switch back; Resources are isolated between clusters, and resources are managed and scheduled based on regions, facilitating database administrators to regulate and restrict the use of database system resources.

The dual-cluster ha solution in the same city further expands the high availability capability of GaussDB. In particular, it provides secure and reliable database services for core financial service scenarios that have higher requirements on performance and stability, protecting the database from disasters and protecting users’ production services.

Technology is endless. In the future, GaussDB(for openGauss) will optimize database HIGH availability technology smarter, more secure, and more efficient to provide customers with more advanced and high-quality database services. More GaussDB (for openGauss) details about please click huawei cloud’s official website: www.huaweicloud.com/product/gau…

Click to follow, the first time to learn about Huawei cloud fresh technology ~