Cloud resource scaling using programmable-network traffic statistics
Some embodiments are associated with systems and methods to provide effective cloud-resource scaling (e.g., for virtual machines and containers). A data center computing system may collect programmable-network real-time traffic statistics associated with enterprise resource planning users. A network and enterprise resource planning load for the enterprise resource planning users may then be predicted (e.g., using a mathematical regression approach) based on an analysis of past programmable-network real-time traffic statistics and current enterprise resource planning user activity. According to some embodiments, the analysis includes time-domain network traffic clustering based on a pre-processing query level categorization. The data center computing system may then dynamically allocate network resources for the enterprise resource planning users based on the predicted network and enterprise resource planning load.
1. A system associated with effective cloud-resource scaling for a data center, comprising:
a data center computing system comprising at least one processor and at least one memory in communication with the at least one processor, wherein the data center computing system is programmed to perform operations including:
collecting programmable-network real-time traffic statistics associated with enterprise resource planning users,
predicting, by a mathematical regression machine learning prediction module of a network control layer control plane, a network and enterprise resource planning load for the enterprise resource planning users based on an analysis of past programmable-network real-time traffic statistics and current enterprise resource planning user activity, and
dynamically allocating network resources of a plurality of virtual machines, each having a plurality of containers, associated with an Infrastructure-as-a-Service (“IaaS”) layer for the enterprise resource planning users based on the predicted network and enterprise resource planning load.
2. The system of claim 1 , wherein enterprise resource planning users and customers access web applications provisioned by a Platform-as-a-Service (“PaaS”) which can run on any IaaS having all hyperscale features.
3. The system of claim 1 , wherein the real-time traffic statistics are associated with an OpenFlow architecture.
4. The system of claim 1 , wherein programmable-network real-time traffic statistics are used to monitor user activity to determine whether or not the user is in a network.
5. The system of claim 4 , wherein the prediction model is associated with clustering based on query hit times, a Central Processing Unit (“CPU”) usage, and a memory usage.
6. The system of claim 4 , wherein the analysis further comprises aggregating clusters for specific users.
7. The system of claim 4 , wherein the programmable-network real-time traffic statistics are associated with timeslot data collected by a monitoring module and a container monitoring module in a network control layer of the data center.
8. A method associated with effective cloud-resource scaling for a data center, comprising:
collecting programmable-network real-time traffic statistics associated with enterprise resource planning users;
predicting, by a mathematical regression machine learning prediction module of a network control layer control plane, a network and enterprise resource planning load for the enterprise resource planning users based on an analysis of past programmable-network real-time traffic statistics and current enterprise resource planning user activity; and
dynamically allocating network resources of a plurality of virtual machines, each having a plurality of containers, associated with an Infrastructure-as-a-Service (“IaaS”) layer for the enterprise resource planning users based on the predicted network and enterprise resource planning load.
9. The method of claim 8 , wherein enterprise resource planning users and customers access web applications provisioned by a Platform-as-a-Service (“PaaS”) which can run on any IaaS having all hyperscale features.
10. The method of claim 8 , wherein the real-time traffic statistics are associated with an OpenFlow architecture.
11. The method of claim 8 , wherein programmable-network real-time traffic statistics are used to monitor user activity to determine whether or not the user is in a network.
12. The method of claim 11 , wherein the prediction module is associated with clustering i-s based on query hit times, a Central Processing Unit (“CPU”) usage, and a memory usage.
13. The method of claim 11 , wherein the analysis further comprises aggregating clusters for specific users and the programmable-network real-time traffic statistics are associated with timeslot data collected by a monitoring module and a container monitoring module in a network control layer of the data center.
14. A non-transitory, machine-readable medium comprising instructions thereon that, when executed by a processor, cause the processor to execute operations, associated with effective cloud-resource scaling for a data center, comprising:
collecting programmable-network real-time traffic statistics associated with enterprise resource planning users;
predicting, by a General Regression Neural Network (“GRNN”) prediction module of a network control layer control plane, a network and enterprise resource planning load for the enterprise resource planning users based on an analysis of past programmable-network real-time traffic statistics and current enterprise resource planning user activity; and
dynamically allocating network resources of a plurality of virtual machines, each having a plurality of containers, associated with an Infrastructure-as-a-Service (“IaaS”) layer for the enterprise resource planning users based on the predicted network and enterprise resource planning load.
15. The medium of claim 14 , wherein the programmable-network real-time traffic statistics further include at least one of memory usage values and Central Processing Unit (“CPU”) utilization values.