Support Questions

Find answers, ask questions, and share your expertise

Single or multiple RDBMS instances for different services

avatar
Expert Contributor

I'm following this tutorial on cluster installation using Ambari

https://docs.hortonworks.com/HDPDocuments/Ambari-2.4.0.1/bk_ambari-installation/content/database_req...

Ambari, Hive, Oozie and Ranger requires RDBMS. Is it a good idea for them to share a single database installation?

Or should I install for each separately? Should separate RDBMS instances is the way to go, for example I choose Postgre 9.3, can I have multiple instances with same version?

1 ACCEPTED SOLUTION

avatar
Master Guru

@J. D. Bacolod

For failure resiliency use separate DB. for ease of operations and maintance use same version.

View solution in original post

2 REPLIES 2

avatar
Master Guru

@J. D. Bacolod

For failure resiliency use separate DB. for ease of operations and maintance use same version.

avatar
Expert Contributor

Assuming that I installed four instances of PostgreSQL 9.3, and what if, for example, the Ranger database fails, it means failure also for HDFS and Hive security (among others). So these components (Ambari, Hive, Oozie, Ranger) are not entirely independent to warrant that a failure in their respective databases means the other will be operating smoothly.

Someone suggested to me to do a single database instance for all four of the services in High Availability mode (master-slave, with warm-standby) or in multi-node cluster, four separate database instances (same distro and version presumably) in High Availability. Although for inexperienced DB admin (like me), this is quite a chore.

As I have read from PostgreSQL documentation, there are a number of solutions for High Availability mode, like Shared Disk Failover, Transaction Log Shipping, etc. What solution did you employ for PostgreSQL HA. Can those who have done this in production cluster share how you did this?

@Sunile Manjee