Question
· Oct 29, 2018

Cache Cluster Environment Backup and Restore

Hi,

I was understanding the cluster configuration supported by the Cache. Have couple of queries around that:

1. In Cache version 2018.2, there is a shrading concept which splits the data of a Master Data server into multiple small data server which store shraded data.

How do we protect (backup and restore) such environment, should we also backup/restore the multiple small data servers along with  Master data server. Any documentation around this will really help.

2. In case of Cache DB Horizontal scaling, we keep adding multiple Application and data servers - to ensure load balancing - for performance.

Do we need to protect (backup and restore) individual Application and data servers separately ? Also is it possible to have two or more data servers in such env. having same instance name and database name.

I am not very pretty clear - on how to protect (backup and restore) above two environments. Request to please help me out here.

Thanks!

Discussion (3)0
Log in or sign up to continue

I'll tackle question 2:

For deciding which backups to take in an ECP environment, it really depends on your setup. In a somewhat typical setup, you'd have the data on the database server, and code on the application servers. In that case, you would definitely want to back up the database server (since that's where the data is), but you may not need to back up the application servers if you have some other repository (such as source control) for your code base. 

I am not quite sure what you mean by having multiple data servers with the same instance name and database name. Can you elaborate? It is certainly possible to mirror your database server in an ECP environment.

Finally, I just want to note that you are talking about sharding, but tagged this as a Caché post. Only InterSystems IRIS supports sharding, not Caché. You probably want to tag this post with IRIS instead of Caché (or both, as your 2 questions really are quite separate).

Thanks for the reply.

I am not quite sure what you mean by having multiple data servers with the same instance name and database name. Can you elaborate? It is certainly possible to mirror your database server in an ECP environment.

  > Is it possible to have multiple Data servers in horizontal scaling,  to offload the data - similar to having multiple Application servers 
If Cache support the above case, what will happen if two data server in it have the same instance name and database name but on different servers. Will that not cause some ambiguity.

My main aim is to protect Cache DB. Thanks for clarifying on sharding part.