mysql - Aurora - MySQL 5.7 - Adaptive_hash_index 配置 - 读取与写入副本

Question

我正在运行 Aurora MySQL 5.7。我已在每个实例的数据库级别将自适应哈希索引设置为 true。 Write 实例设置为 true，但只读副本仍设置为 false。不知道为什么。难道这不能改变吗？

Answer 1

这似乎是 Aurora 的已知且已验证(未记录)的限制。根据 Percona 博客文章“Adaptive Hash Index on AWS Aurora”，AWS 工程师已确认无法在 Reader 节点上启用自适应哈希索引。如果您有需要 AHI 的工作负载，则需要在 Writer 节点或单独的 MySQL 副本上运行它们。

事实上，您仍然可以在 Reader 实例上修改此参数组参数(该参数无效)，这似乎是他们正在调查的一个错误。

另请参阅this Aug 2015 AWS forum post by Stefano@AWS这证实了这个限制:

Question on performance of queries with joins across large datasets. As some have pointed out, enabling innodb_adaptive_hash_index can improve this class of query. The parameter is disabled by default in Aurora as it can be detrimental on performance intensive mixed read/write workloads, but can be re-enabled at any time via parameter group. I can confirm that at the moment the parameter is only honored on the master instance of the cluster and not by read replicas. It is possible to workaround the limitation by creating a separate Aurora instance and enabling binlog replication, which is something we support. This is an area where we have improvements planned and are currently working on.