Running multiple databases concurrently

[verticanewbee]

1. Can a cluster run multiple databases in one particular instant?

2. Can a node in a particular cluster run multiple databases?

3. Can a same database be loaded in multiple clusters and worked on concurrently?

I wanted to know whether the above scenarios are possible or not, and with a detailed explanation please.

Thanks in advance...

[nnani]

Hello,

1. Can a cluster run multiple databases in one particular instant?

No, In Vertica, you can only run one database at a time in an instant, So If you wish to use two database at a time, stop one and start the new one.
More then one database cannot run simultaneously in an instant.

2. Can a node in a particular cluster run multiple databases?

No, all nodes are in sync, so I believe you can have only one initiator node at a time.
In case If you tried this, already have a database running and try starting a database from another node, you will get a promt that one database is already running. ( try this )

3. Can a same database be loaded in multiple clusters and worked on concurrently?

You can work concurrently on different clusters using the vsql client but on different instances. One instance for a cluster and another for different cluster
You need to configure your vsql client, if working on a remote machine and parametrize it with the proper credentials while connnecting to different clusters.
Regarding loading the same database, Set up identical database on different clusters and then use the "Export to Vertica" feature to keep both database in sync ( this can be a hectic process to do this every now and then )


Hope this helps

[NorbertKrupa]

You can have multiple databases. Each database would need dedicated nodes. With a 4 node cluster:

DB1 on node1, node2
DB2 on node3, node4

The way Vertica is designed is intended for a single database instance. Vertica falls under the MPP (Massively Parallel Processing) category. Multiple databases would be competing for resources on across nodes in a cluster. The parallel design enables the distribution of storage and workload across the nodes. In my opinion, the best design is to logically create your schemas like you would databases.

[pborne]

It is technically possible to run multiple instances of Vertica on the same cluster at the same time. They would use different port numbers and you have to start them up "by hand" on the command line, not by using admintools. However, this is not supported and obviously not recommended to use in production. This is a feature that is mostly used when testing or developing. Each Vertica instance runs with the assumption that is "owns" the hardware it runs on. Obviously, if you try to run massive queries at the same time on different instances, you will run out of memory very quickly.

For what you need (we don't know all the details) you should either run separate clusters, or, a single cluster with a single instance but use multiple schemas. This way, the engine is aware of every query running and you can use resource pools to control resource consumption.