ApsaraDB for HBase provides three basic read and write APIs: read (GET, SCAN), write (PUT). In actual use, there may be instances where you want to perform these operations at the same time without interfering with each other. However, read/write splitting is not enabled by default. To achieve this effect, perform the following steps:

  • If your business handles high volumes of reads and writes and you want to prioritize reads over writes, we recommend that you implement read/write splitting.
  • If your business handles a large volume of SCAN and GET requests, it is desired that SCAN requests do not affect the performance of GET requests.

Related configurations:

  • hbase.ipc.server.callqueue.read.ratio
  • hbase.ipc.server.callqueue.scan.ratio

Description:

  • Setting the hbase.ipc.server.callqueue.read.ratio to 0.5 indicates that 50% of the threads are used for read requests.
  • If you also set hbase.ipc.server.callqueue.scan.ratio to 0.5, 50% of the read threads are used for SCAN requests, which means that 25% of the total threads are used for SCAN requests.

Procedure

  • Click the cluster in the ApsaraDB for HBase console and select Parameter Configuration.
  • Modify the configuration based on the read/write status of the business.
  • The modifications take effect only after the cluster is restarted. Restarting the cluster will not cause major business interruptions, but may cause network jitters. We recommend that you restart the cluster during off-peak hours to prevent any potential negative effects.set_value

Configure the preceding parameters based on your business needs. The parameters are not specified by default. This indicates that read and write operations share threads.