Prescriptive analytics based committed compute reservation stack for cloud computing resource scheduling

Patent metadata

Field	Value
Publication number	US-10922141-B2
Application number	US-201815922659-A
Country	US
Kind code	B2
Filing date	Mar 15, 2018
Priority date	Dec 11, 2017
Publication date	Feb 16, 2021
Grant date	Feb 16, 2021

Abstract

A multi-layer committed compute reservation stack may generate prescriptive reservation matrices for controlling static reservation for computing resources. A transformation layer of the committed compute reservation stack may generate a time-mapping based on historical utilization and tagging data. An iterative analysis layer may determine a consumption-constrained committed compute state of a distribution of static reservation and dynamic requisition that achieves one or more consumption efficiency goals. Once the consumption-constrained committed compute state is determined, the prescriptive engine layer may generate a reservation matrix that may be used to control computing resource static reservation prescriptively.

First claim

What is claimed is:

1. A system comprising:
network interface circuitry configured to:
receive historical resource utilization data for a set of virtual machines;
receive consumption metric data for the set of virtual machines;
receive tagging data defining a functional grouping for a first virtual machine of the set of virtual machines;
send a reservation matrix to a host interface configured to control static reservation and dynamic requisition for at least the first virtual machine, the reservation matrix including one or more requests for one or more committed compute virtual machines of the set of virtual machines;

reservation circuitry in data communication with the network interface circuitry, the reservation circuitry configured to execute a committed compute reservation (CCR) stack,
the CCR stack comprising:
a data staging layer;
an input layer;
a transformation layer;
an iterative analysis layer; and
a prescriptive engine layer;

the CCR stack configured to:
obtain, via a data control tool at the input layer, the historical utilization data, the consumption metric data, and the tagging data;
store, at the data staging layer, the historical utilization data, the consumption metric data, and the tagging data;
access, at the transformation layer, the historical utilization data and the tagging data via a memory resource provided by the data staging layer;
process, at the transformation layer, the historical utilization data and the tagging data to generate a time-mapping of active periods for virtual machines within the functional grouping;
store, via operation at the data staging layer, the time-mapping;
obtain, at the iterative analysis layer, a current committed compute (CC) state for the set of virtual machines, the current CC state detailing one or more pre-analysis committed compute states for one or more virtual machines of the set of virtual machines;
access, at the iterative analysis layer, the consumption metric data;
within an analysis space of CC states, use boundary conditions to determine a search space for a non-linear search, the search space around the current CC state, where a disallowed CC state is outside the search space but within the analysis space;
based on the time-mapping, the consumption metric data, and the previous CC state, determine, responsive to a non-linear search iteration of the non-linear search, a consumption-constrained CC state comprising a static reservation prescription for at least the first virtual machine;
store, via operation at the data staging layer, the consumption-constrained CC state;
access the consumption-constrained CC state at the prescriptive engine layer;
analyze the consumption-constrained CC state responsive to a feedback history, the feedback history based on commands from a committed compute (CC) control interface;
responsive to analyzing the consumption-constrained CC state, generate, at the prescriptive engine layer, a committed compute window (CC-window) presentation for the CC control interface; and
after generation of the CC-window presentation, generate the reservation matrix based on at least the consumption-constrained CC state and the feedback history.

2. The system of claim 1 , where the CCR stack is configured to obtain the current CC state based on an initial CC state describing a pre-analysis distribution of static reservation and dynamic requisition.

3. The system of claim 1 , where the CCR stack is configured to, at the iterative analysis layer, iteratively obtain the consumption-constrained CC state by obtaining the current CC state by using the non-linear search around a previous CC state.

4. The system of claim 1 , where the CC-window presentation comprises consumption metric data for the consumption-constrained CC state.

5. The system of claim 4 , where the CC-window presentation is configured to present the consumption metric data for the consumption-constrained CC state alongside consumption metric data for multiple selectable CC state options.

6. The system of claim 1 , where the CCR stack is configured to determine consumption-constrained CC state by applying a nonlinear optimization algorithm.

7. The system of claim 6 , where the nonlinear optimization algorithm comprises a constrained optimization by linear approximation (COBYLA) routine.

8. The system of claim 1 , where the historical utilization data comprises resource allocation history, activation history data, reservation history data, committed-use history data, expenditure report data, processor activity, memory usage history, computing cycles consumption, data throughput, temporal usage cycle data, or any combination thereof.

9. The system of claim 1 , where tagging data comprises virtual machine provisioning data, virtual machine functional group definitions, project-specific allocation data, availability zone data, operating system data, quality of service data, or any combination thereof.

10. The system of claim 1 , where the CCR stack is configured to generate the reservation matrix based further on an input selection received from the CC control interface responsive to the CC-window presentation.

11. The system of claim 1 , where the CCR stack is configured to and responsive to the feedback history and at the prescriptive engine layer, refine a consumption savings threshold for generation of the reservation matrix.

12. A method including:
at network interface circuitry:
receiving historical resource utilization data for a set of virtual machines;
receiving consumption metric data for the set of virtual machines; and
receiving tagging data defining a functional grouping for a first virtual machine of the set of virtual machines;

at reservation circuitry in data communication with the network interface circuitry, the reservation circuitry configured to execute a committed compute reservation (CCR) stack:
obtaining, via a data control tool at an input layer of the CCR stack, the historical utilization data, the consumption metric data, and the tagging data;
storing, at a data staging layer of the CCR stack, the historical utilization data, the consumption metric data, and the tagging data;
accessing, at a transformation layer of the CCR stack, the historical utilization data and the tagging data via a memory resource provided by the data staging layer;
processing, at the transformation layer, the historical utilization data and the tagging data to generate a time-mapping of active periods for virtual machines within the functional grouping;
storing, via operation at the data staging layer, the time-mapping;
at an iterative analysis layer of the CCR stack:
obtaining a current committed compute (CC) state for the set of virtual machines, the current CC state detailing one or more pre-analysis committed compute states for one or more virtual machines of the set of virtual machines;
accessing the consumption metric data;
within an analysis space of CC states, based on boundary conditions, determining a search space for a non-linear search, the search space around the current CC state, where a disallowed CC state is outside the search space but within the analysis space; and
based on the time-mapping and the consumption metric data, determining, responsive to a non-linear search iteration of the non-linear search, a consumption-constrained CC state comprising a static reservation prescription for at least the first virtual machine;

storing, via operation at the data staging layer, the consumption-constrained CC state;
at a prescriptive engine layer of the CCR stack:
accessing the consumption-constrained CC state;
analyzing the consumption-constrained CC state responsive

Assignees

• Accenture Global Solutions Ltd

Inventors

• Srinivasan Madhan Kumar
• Purushothaman Arun
• Kolachalam Manish Sharma
• Eisenstein Michael S

Classifications

• G06F9/451

  Execution arrangements for user interfaces · CPC title

• G06F9/5077 Primary

  Logical partitioning of resources; Management or configuration of virtualized resources (specific details on emulation or internal functioning of virtual machines G06F9/455) · CPC title

• G06F2209/5014

  Reservation · CPC title

• G06F9/45533

  Hypervisors; Virtual machine monitors · CPC title