gridgain / gridgain-old

267 stars 85 forks source link

Self Tuning Grid #64

Closed sirinath closed 10 years ago

sirinath commented 10 years ago

It is possible to have a self tuning grid where all configuration, data and code affinities, nodes and deployments are self tuned.

(Simplification of following: https://github.com/gridgain/gridgain/issues/56)

Also work with cloud management system (https://github.com/gridgain/gridgain/issues/63) to similarly optimise cloud deployments without mediation, in this case automatically adding / removing new nodes to optimise workload, throughput and cost.

dsetrakyan commented 10 years ago

This would be default configuration.

sirinath commented 10 years ago

OK.

sirinath commented 10 years ago

Will the cloud tuning part be included when you have the Cloud Cartridges?

dsetrakyan commented 10 years ago

Yes, this will be default cloud configuration.

sirinath commented 10 years ago

May be you did not get what I meant. Let me rephrase. You have to option to add and remove nodes in a cloud. It is not a static grid with finite number of nodes. Will GG be able to manage the cloud resources in the best possible way by adding and removing nodes to optimise cost while maintaining throughput and performance. Also cloud cost and plans differ from provider to provider and this needs to be taken into account perhaps through configuration.

Basically you will have to actively manage the cloud deployment and resources here.

Just to make sure we are in the same page regarding the upcoming cloud features. I am not sure whether the current default configuration has this feature to manage the cloud also. My current understanding is when you have the cloud features you will add the needed features to actively optimise the GG deployment for the cloud by also actively managing the cloud deployment like number of nodes, data transfer, data storage, etc.