Intent-based orchestration in heterogenous compute platforms

Patent metadata

Field	Value
Publication number	US-12299113-B2
Application number	US-202117561061-A
Country	US
Kind code	B2
Filing date	Dec 23, 2021
Priority date	Nov 16, 2021
Publication date	May 13, 2025
Grant date	May 13, 2025

Abstract

Various systems and methods for implementing intent-based orchestration in heterogenous compute platforms are described herein. An orchestration system is configured to: receive, at the orchestration system, a workload request for a workload, the workload request including an intent-based service level objective (SLO); generate rules for resource allocation based on the workload request; generate a deployment plan using the rules for resource allocation and the intent-based SLO; deploy the workload using the deployment plan; monitor performance of the workload using real-time telemetry; and modify the rules for resource allocation and the deployment plan based on the real-time telemetry.

First claim

What is claimed is:

1. An orchestration system for implementing intent-based orchestration in heterogenous compute platforms, comprising:
a processor; and
memory to store instructions, which when executed by the processor, cause the orchestration system to:
receive, at the orchestration system, a workload request for a workload, the workload request including an intent-based service level objective (SLO);
generate rules for resource allocation based on the workload request;
generate a deployment plan using the rules for resource allocation and the intent-based SLO;
deploy the workload using the deployment plan;
monitor performance of the workload using real-time telemetry; and
modify the rules for resource allocation and the deployment plan based on the real-time telemetry.

2. The orchestration system of claim 1 , wherein the rules for resource allocation are generated based on lookup tables.

3. The orchestration system of claim 2 , wherein the lookup tables match a type of workload to resources from a pool of resources.

4. The orchestration system of claim 1 , wherein the deployment plan includes a reservation of resources for the workload.

5. The orchestration system of claim 1 , wherein the deployment plan includes a predictive instantiation of resources for the workload.

6. The orchestration system of claim 1 , wherein the deployment plan is generated using a machine learning model.

7. The orchestration system of claim 1 , wherein the deployment plan is used to reconfigure hardware resources to service the workload request.

8. The orchestration system of claim 7 , wherein to reconfigure hardware, the deployment plan is used to program a programmable hardware unit.

9. The orchestration system of claim 8 , wherein the programmable hardware unit is a field-programmable gate array (FPGA).

10. The orchestration system of claim 1 , wherein the deployment plan includes a specification of a type of core to use.

11. The orchestration system of claim 10 , wherein multiple different types of cores are disposed on the same die.

12. The orchestration system of claim 1 , wherein the deployment plan includes a specification of a node instance to use.

13. The orchestration system of claim 12 , wherein the node instance is provided by an owner of the workload.

14. The orchestration system of claim 13 , wherein the owner of the workload provides nodes that are permitted for the workload and nodes that are not permitted for the workload.

15. The orchestration system of claim 1 , wherein the deployment plan includes a security mechanism to isolate a workload container from other containers executing on a node, the workload container associated with the workload and the other containers not associated with the workload.

16. The orchestration system of claim 15 , wherein the security mechanism includes dynamically assigned keys to encrypt container processing at the node.

17. The orchestration system of claim 1 , wherein to modify the rules for resource allocation and the deployment plan, the orchestration system is to:
profile execution of the workload; and
migrate a portion of the workload to a different compute node to improve compliance with the intent-based SLO.

18. The orchestration system of claim 1 , wherein to monitor performance of the workload, a pseudo-sidecar that is deployed with the workload and is executing on a compute node, is configured to collect and report the real-time telemetry.

19. The orchestration system of claim 18 , wherein the pseudo-sidecar is one of a plurality of pseudo-sidecars deployed with corresponding containers that are configured to execute the workload.

20. The orchestration system of claim 19 , wherein a node agent executing at the compute node is used to monitor performance of the workload and modify the rules for resource allocation.

21. At least one machine-readable medium including instructions for implementing intent-based orchestration in heterogenous compute platforms, which when executed by a machine, cause the machine to perform operations comprising:
receiving, at an orchestration system, a workload request for a workload, the workload request including an intent-based service level objective (SLO);
generating rules for resource allocation based on the workload request;
generating a deployment plan using the rules for resource allocation and the intent-based SLO;
deploying the workload using the deployment plan;
monitoring performance of the workload using real-time telemetry; and
modifying the rules for resource allocation and the deployment plan based on the real-time telemetry.

22. The machine-readable medium of claim 21 , wherein modifying the rules for resource allocation and the deployment plan, includes:
profiling execution of the workload; and
migrating a portion of the workload to a different compute node to improve compliance with the intent-based SLO.

23. The machine-readable medium of claim 21 , wherein to monitor performance of the workload, a pseudo-sidecar that is deployed with the workload and is executing on a compute node, is configured to collect and report the real-time telemetry.

24. The machine-readable medium of claim 23 , wherein the pseudo-sidecar is one of a plurality of pseudo-sidecars deployed with corresponding containers that are configured to execute the workload.

25. The machine-readable medium of claim 24 , wherein a node agent executing at the compute node is used to monitor performance of the workload and modify the rules for resource allocation.

Assignees

• Intel Corp

Inventors

• Metsch Thijs
• Balle Susanne M
• Koeberl Patrick
• Li Bin
• Yarvis Mark
• Hoban Adrian
• Doshi Kshitij Arun
• Guim Bernat Francesc
• Martinez-Spessot Cesar
• Agerstam Mats Gustav
• Oliver Dario Nicolas
• Carranza Marcos E
• Browne John J
• Ylinen Mikko
• Cremins David

Classifications

• H04L47/83

  based on usage prediction · CPC title

• H04L67/52

  specially adapted for the location of the user terminal · CPC title

• G06Q10/087

  Inventory or stock management, e.g. order filling, procurement or balancing against orders · CPC title

• G06F9/44505

  Configuring for program initiating, e.g. using registry, configuration files · CPC title

• G06F9/3875

  Pipelining a single stage, e.g. superpipelining · CPC title