Ask Your Question
0

Why ODL is having 2 stores ( Config & Operational )

asked 2016-09-03 06:35:11 -0700

kspviswa gravatar image

Hello,

I would like to know what is the notion / rationale behind maintaining 2 stores. While I understand config store persist incoming request and operational store reflect the network state, I want to understand what is the problem is maintaining both state in same store.

i.e For example, persist the incoming request in a store, try to apply the request in network. If the network operation fails, then revert the store to previous state ( roll back ) else, commit the transaction in the store.

This way, we can have the consistency with single store. If that's the case, why ODL maintains 2 store?

BR,

Viswa

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
1

answered 2016-09-05 00:21:03 -0700

The config and operational data have different characteristics is terms of performance and persistence needs. While config data is mostly "intent", operational data is mostly "state". Config data is persisted, and because of that, the config data store may be slower. Operational data is updated more frequently, and is such needs higher performance, but no persistence. Please note that the data stored are clustered, and maintaining a distributed data store implies certain trade-offs to be made. Knowing the data characteristics makes it easier to optimize for certain type of data.

edit flag offensive delete publish link more
Login/Signup to Answer

Stats

Asked: 2016-09-03 06:35:11 -0700

Seen: 55 times

Last updated: Sep 05 '16