Overview
下载
聚焦模式
字号
Cross-Region Message Replication Scenarios
Cross-Region Cold Disaster Recovery
Cross-Region Cold Disaster Recovery is a low-cost disaster recovery solution designed for region-level failures: It involves deploying a standby cluster in a different region. Under normal conditions, the primary cluster handles the business load, while the standby cluster periodically synchronizes metadata such as Topics, subscriptions, and permissions to maintain a ready-to-take-over state. This approach enables low-specification deployment and controlled costs. When an exception occurs in the primary region, users can perform an access point domain name resolution switchover in the console to quickly divert business traffic to the standby cluster. Clients do not need to change their configurations, and the standby cluster supports elastic scaling to handle the post-switchover peak traffic. After the primary region recovers, the accumulated data generated on the standby side can be written back to the primary cluster as needed. This completes the disaster recovery cycle, ensuring business continuity and data integrity.
Cross-Region Hot Disaster Recovery
Cross-Region Hot Disaster Recovery is a high-availability disaster recovery solution designed for region-level failures: It deploys clusters across two regions based on GEO Replication and uses Replicator to synchronize metadata and messages in real time, thereby achieving multi-region message replication. This solution is suitable for scenarios such as cross-border message synchronization. It also supports cross-cluster metadata management, enabling a single modification to be synchronized across all regions, which facilitates Ops and operations. For the synchronization mechanism, it provides real-time synchronization to ensure configuration consistency across clusters and performs regular reconciliation as a safety net to verify data consistency. Furthermore, it adheres to a clear read/write sequence: When metadata is added, the target cluster is written first, followed by the source cluster; when metadata is deleted, the source cluster is written first, followed by the target cluster. This further enhances business continuity assurance.
Cross-Region Message Replication for TDMQ for Apache Pulsar
Based on the two cross-region cold and hot disaster recovery scenarios described above, TDMQ for Apache Pulsar pro clusters support two types of cross-region message replication:
Message-Level Cross-Region Replication: This feature is primarily used for global data archiving, multi-center data sharing, and cross-region consumption scenarios. It synchronizes the actual message content across regions to the disaster recovery cluster in real time. This makes it well-suited for scenarios that require aggregating data streams from multiple regions nationwide or globally into a single data center for unified archiving and centralized big data analysis, or for enabling cross-region systems to consume the same batch of business data. For details, see Message-Level Cross-Region Replication.
Metadata-Level Cross-Region Replication: This feature primarily addresses scenarios requiring cross-region disaster recovery for users' core business. It only synchronizes "configuration structures" such as namespaces, topics, and subscription relationships between clusters in both regions, without replicating specific message content (thus incurring no cross-region traffic fees). It is suitable for enabling business clients to smoothly switch to the disaster recovery cluster with one click when the primary cluster encounters extreme physical disasters, quickly re-establish connections, and continue sending and receiving messages, thereby ensuring business continuity. For details, see Metadata-Level Cross-Region Replication.
Billing Overview
Message-level cross-region replication tasks incur fees, while metadata-level cross-region replication tasks do not incur fees. The console provides the corresponding monitoring metrics to facilitate observation.
For detailed billing overview about message-level cross-region replication, see Billing Overview of Pro Clusters.
Use Limits
1. Supported Scope
This feature is supported only by pro clusters - fixed/elastic storage.
2. Supported Regions
Shanghai, Nanjing, Guangzhou, and Shenzhen Finance.
3. Link Replication Directions
Currently, both message-level cross-region replication and metadata-level cross-region replication are unidirectional.
4. Replication Objects
Only one synchronization task, including tasks at higher levels, can be created for the same object.
Do not delete topics in a synchronization link. You need to first delete the synchronization link.
5. Overwriting
For existing resources with the same name but different configurations, such as the number of partitions and TTL, the configuration of the target cluster will not be overwritten after a replication task is created.
文档反馈