System and method for resource placement across clouds for data intensive workloads

• US 10,705,882 B2
• Filed: 12/21/2017
• Issued: 07/07/2020
• Est. Priority Date: 12/21/2017
• Status: Active Grant

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Inventors: Dutta, Debojyoti, Huang, Xinyuan
• Primary Examiner(s): Sall, El Hadji M
• Application Number:
  US15/850,230
• Publication Number:
  US 20190196879A1
• Time in Patent Office: 929 Days
• CPC Class Codes: G06F 9/5038 considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration scheduling strategies G06F9/4881 and subgroups
• CPC Class Codes: G06F 9/5044 considering hardware capabilities
• CPC Class Codes: G06F 9/505 considering the load
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system
• CPC Class Codes: H04L 41/0803 Configuration setting

Abstract

Systems, methods, computer-readable media are disclosed for determining a point of delivery (POD) device or network component on a cloud for workload and resource placement in a multi-cloud environment. A method includes determining a first amount of data for transitioning from performing a first function on input data to performing a second function on a first outcome of the first function; determining a second amount of data for transitioning from performing the second function on the first outcome to performing a third function on a second outcome of the second function; determining a processing capacity for each of one or more network nodes on which the first function and the third function are implemented; and selecting the network node for implementing the second function based on the first amount of data, the second amount of data, and the processing capacity for each of the network nodes.

Claims

1. A computer-implemented method for selecting a network to implement a serverless function, the method comprising:

determining, by one or more processors of a network controller, a first throughput based on a first amount of data for transitioning from performing a first serverless function on input data within a source network to performing a second serverless function on a first outcome of the first serverless function;

determining, by the one or more processors, a second throughput based on a second amount of data for transitioning from performing the second serverless function on the first outcome to performing a third serverless function on a second outcome of the second serverless function within a sink network;

determining a processing capacity for each of one or more networks on which the first serverless function and the third serverless function are implemented; and

selecting a network for implementing the second serverless function based on the first throughput, the second throughput, and the processing capacity for each of the networks.

2. The method of claim 1, wherein the first serverless function and the second serverless function are implemented on the sink network.

3. The method of claim 2, further comprising:

determining the processing capacity of the source network is equal to or greater than the first throughput and the second throughput.

4. The method of claim 1, further comprising:

implementing the second serverless function on the sink network if the processing capacity of the source network is less than the first throughput and the second throughput.

5. The method of claim 4, further comprising:

determining a second processing capacity of the sink network is greater than the first throughput and the second throughput and a cost associated with transitioning data corresponding to the first outcome and the second serverless function from the source network to the sink network is minimized among all costs associated with transitioning the data corresponding to the first outcome and the second serverless function from the source network to any other available network.

6. The method of claim 1, wherein the first serverless function and the third serverless function are implemented on the source network and the sink network, respectively, and the method further comprises:

determining a first processing capacity of the source network;

determining a second processing capacity of the sink network;

implementing the second serverless function on the source network if the first processing capacity of the source network is equal to or greater than the first throughput and the second throughput;

implementing the second serverless function on the sink network if,

(1) the first processing capacity of the source network is less than the first throughput and the second throughput, and

(2) the second processing capacity of the sink network is equal to or greater than the first throughput and the second throughput; and

implementing the second serverless function on a third network if,

(1) the first processing capacity of the source network is less than the first throughput and the second throughput,

(2) the second processing capacity of the sink network is less than the first throughput and the second throughput,

(3) a third processing capacity of the third network is equal to or greater than the first throughput and the second throughput, and

(4) a cost associated with transitioning data corresponding to the first outcome and the second serverless function from the source network to the third network is minimized among all costs associated with transitioning the data corresponding to the first outcome and the second serverless function from the source network to any other available network.

7. The method of claim 1, further comprising:

receiving an indication of a plurality of data segments to be processed according to the first serverless function, the second serverless function, and the third serverless function at a predetermined time;

at one or more of a plurality of target times prior to the predetermined time,

performing the first serverless function, the second serverless function, and the third serverless function on available ones of the plurality of data segments to yield first processed data; and

storing the first processed data.

8. The method of claim 7, further comprising:

determining, at the predetermined time, if the plurality of data segments are unchanged;

if the plurality of data segments are unchanged,

performing the first serverless function, the second serverless function, and the third serverless function on remaining ones of the plurality of data segments received between a last one of the plurality of target times and the predetermined time to yield second processed data;

storing the second processed data; and

combining the first processed data and the second processed data; and

if the plurality of data segments are changed,

discarding the first processed data;

performing the first serverless function, the second serverless function, and the third serverless function on a complete set of the plurality of data segments to yield third processed data; and

storing the third processed data.

9. A system for selecting a network to implement a serverless function, the system comprising:

memory configured to store computer-readable instructions therein; and

one or more processors configured to execute the computer-readable instructions to;

determine a first throughput based on a first amount of data for transitioning from performing a first serverless function on input data within a source network to performing a second serverless function on a first outcome of the first serverless function;

determine a second throughput based on a second amount of data for transitioning from performing the second serverless function on the first outcome to performing a third serverless function on a second outcome of the second serverless function within a sink network;

determine a processing capacity for each of one or more networks on which the first serverless function and the third serverless function are implemented; and

select a network for implementing the second serverless function based on the first throughput, the second throughput, and the processing capacity for each of the networks.

10. The system of claim 9, wherein the first serverless function and the third serverless function are implemented on the source network, and the one or more processors are configured to execute the computer-readable instructions to further:

determine if the processing capacity of the source network is equal to or greater than the first throughput and the second throughput; and

implement the second serverless function on the sink network if the processing capacity of the source network is less than the first throughput and the second throughput.

11. The system of claim 10, wherein the one or more processors are configured to execute the computer-readable instructions to further:

determine a second processing capacity of the sink network is greater than the first throughput and the second throughput and a cost associated with transitioning data corresponding to the first outcome and the second serverless function from the source network to the sink network is minimized among all costs associated with transitioning the data corresponding to the first outcome and the second serverless function from the source network to any other available network.

12. The system of claim 10, wherein the source network and the sink network are associated with different cloud environments.

13. The system of claim 9, wherein the first serverless function and the second serverless function are implemented on the source network and the sink network, respectively, and the one or more processors are configured to execute the computer-readable instructions to further:

determine a first processing capacity of the source network;

determine a second processing capacity of the sink network;

implement the second serverless function on the source network if the first processing capacity of the source network is equal to or greater than the first throughput and the second throughput;

implement the second serverless function on the sink network if,

(1) the first processing capacity of the source network is less than the first throughput and the second throughput, and

(2) the second processing capacity of the sink network is equal to or greater than the first throughput and the second throughput; and

implement the second serverless function on a third network if,

(1) the first processing capacity of the source network is less than the first throughput and the second throughput,

(2) the second processing capacity of the sink network is less than the first throughput and the second throughput,

(3) a third processing capacity of the third network is equal to or greater than the first throughput and the second throughput, and

(4) a cost associated with transitioning data corresponding to the first outcome and the second serverless function from the source network to the third network is minimized among all costs associated with transitioning the data corresponding to the first outcome and the second serverless function from the source network to any other available network.

14. The system of claim 9, wherein the one or more processors are configured to execute the computer-readable instructions to further:

receive an indication of a plurality of data segments to be processed according to the first serverless function, the second serverless function, and the third serverless function at a predetermined time;

at one or more of a plurality of target times prior to the predetermined time,

perform the first serverless function, the second serverless function, and the third serverless function on available ones of the plurality of data segments to yield first processed data; and

store the first processed data.

15. The system of claim 14, wherein the one or more processors are configured to execute the computer-readable instructions to further:

determine, at the predetermined time, if the plurality of data segments are unchanged;

if the plurality of data segments are unchanged,

perform the first serverless function, the second serverless function, and the third serverless function on remaining ones of the plurality of data segments received between a last one of the plurality of target times and the predetermined time to yield second processed data;

store the second processed data; and

combine the first processed data and the second processed data; and

if the plurality of data segments are changed,

discard the first processed data;

perform the first serverless function, the second serverless function, and the third serverless function on a complete set of the plurality of data segments to yield third processed data; and

store the third processed data.

16. One or more non-transitory computer-readable medium having computer-readable instructions stored therein, which when executed by one or more processors, cause the one or more processors to:

determine a first throughput based on a first amount of data for transitioning from performing a first serverless function on input data within a source network to performing a second serverless function on a first outcome of the first serverless function;

determine a second throughput based on a second amount of data for transitioning from performing the second serverless function on the first outcome to performing a third serverless function on a second outcome of the second serverless function within a sink network;

determine a processing capacity for each of one or more networks on which the first serverless function and the third serverless function are implemented; and

select a network for implementing the second serverless function based on the first throughput, the second throughput, and the processing capacity for each of the networks.

17. The one or more non-transitory computer-readable medium of claim 16, wherein the first serverless function and the third serverless function are implemented on the source network, and the one or more processors are configured to execute the computer-readable instructions to further:

determine if a first processing capacity of the source network is equal to or greater than the first throughput and the second throughput; and

implement the second serverless function on the sink network if the processing capacity of the source network is less than the first throughput and the second throughput, a second processing capacity of the sink network is greater than the first throughput and the second throughput, and a cost associated with transitioning data corresponding to the first outcome and the second serverless function from the source network to the sink network is minimized among all costs associated with transitioning the data corresponding to the first outcome and the second serverless function from the source network to any other available network.

18. The one or more non-transitory computer-readable medium of claim 16, wherein the first serverless function and the second serverless function are implemented on the source network and the sink network, respectively, and the one or more processors are configured to execute the computer-readable instructions to further:

a first processing capacity of the source network;

a second processing capacity of the sink network;

implement the second serverless function on the source network if the first processing capacity of the source network is equal to or greater than the first throughput and the second throughput;

implement the second serverless function on the sink network if,

(1) the first processing capacity of the source network is less than the first throughput and the second throughput, and

(2) the second processing capacity of the sink network is equal to or greater than the first throughput and the second throughput; and

implement the second serverless function on a third network if,

(1) the first processing capacity of the source network is less than the first throughput and the second throughput,

(2) the second processing capacity of the sink network is less than the first throughput and the second throughput,

(3) a third processing capacity of the third network is equal to or greater than the first throughput and the second throughput, and

(4) a cost associated with transitioning data corresponding to the first outcome and the second serverless function from the source network to the third network is minimized among all costs associated with transitioning the data corresponding to the first outcome and the second serverless function from the source network to any other available network node.

19. The one or more non-transitory computer-readable medium of claim 16, wherein the one or more processors are configured to execute the computer-readable instructions to further:

receive an indication of a plurality of data segments to be processed according to the first serverless function, the second serverless function, and the third serverless function at a predetermined time;

at one or more of a plurality of target times prior to the predetermined time,

perform the first serverless function, the second serverless function, and the third serverless function on available ones of the plurality of data segments to yield first processed data; and

store the first processed data.

20. The one or more non-transitory computer-readable medium of claim 19, wherein the one or more processors are configured to execute the computer-readable instructions to further:

determine, at the predetermined time, if the plurality of data segments are unchanged;

if the plurality of data segments are unchanged,

perform the first serverless function, the second serverless function, and the third serverless function on remaining ones of the plurality of data segments received between a last one of the plurality of target times and the predetermined time to yield second processed data;

store the second processed data; and

combine the first processed data and the second processed data; and

if the plurality of data segments are changed,

discard the first processed data;

perform the first serverless function, the second serverless function, and the third serverless function on a complete set of the plurality of data segments to yield third processed data; and

store the third processed data.