| CPC G06F 9/5077 (2013.01) [G06F 9/5027 (2013.01)] | 3 Claims |
|
1. A cluster node recommendation system comprising:
an input unit to which user selection information is input from a user, the user selection information including at least one of a cloud vendor, an Information Technology (IT) resource size, and a free resource size;
a resource requirement checking unit configured to check resource requirements of a designated application;
an artificial intelligence module to which the user selection information input to the input unit and the resource requirements of the application checked by the resource requirement checking unit are input and which outputs a node configuration;
a validity verification unit configured to verify validity by arranging a container in which the application is executed, in the node configuration output by the artificial intelligence module;
a recommendation result output unit configured to recommend a final node configuration in which validity verification by using the validity verification unit is successful, to the user;
a recommendation satisfaction collection unit configured to receive an approval signal of the final node configuration recommended by the recommendation result output unit and then to match the recommended final node configuration, identification information of the application, whether or not approval is made, with each other and to store a result of matching in a result processing list;
an output correction unit configured to direct output of a new node configuration to the artificial intelligence module when a disapproval signal is received by the recommendation satisfaction collection unit;
wherein the validity verification unit checks the result processing list and performs validity verification when the node configuration output by the artificial intelligence module and the application identification information are matched with each other; and
wherein the validity verification unit checks the result processing list and determines that validity has failed when a user disapproval ratio of the node configuration output by the artificial intelligence module and the application identification information are matched with each other, exceeds a preset ratio; and
the artificial intelligence module performs machine learning periodically by using previously stored learning data,
when an amount of the previously stored learning data is less than a preset value, the cluster node recommendation system further comprises a learning data expansion unit configured to expand virtual learning data by using the learning data, and
the learning data expansion unit determines expansion weights for each learning data based on information stored in the result processing list.
|