Management of serverless function deployments in computing networks
Systems, methods, computer-readable media are disclosed for influencing serverless function placement across hosts within a network. A method includes receiving a notification from a network component, the notification indicating a performance bottleneck in association with one or more instances of a serverless function being executed at one or more hosts of a network; initiating at least one additional instance of the serverless function in response to the performance bottleneck; and sending a message to the network component identifying the at least one additional instance of the serverless function, the network component directing network traffic based on the message.
1. A method comprising:
receiving, at an orchestrator, a first notification from a network component, the first notification indicating a performance bottleneck in association with one or more instances of a serverless function being executed atone or more hosts of a network;
initiating, at the orchestrator, at least one additional instance of the serverless function in response to the performance bottleneck;
sending, at the orchestrator, a first message to the network component identifying the at least one additional instance of the serverless function, the network component directing network traffic based on the first message;
determining, at the orchestrator, whether the at least one additional instance of the serverless function is still needed;
in response to determining the at least one additional instance of the serverless function is not still needed, sending, at the orchestrator, a second message to the network component and host handling the at least one additional instance of the serverless function, the second message informing the network component and host the at least one additional instance of the serverless function is to be destroyed; and
receiving, at the orchestrator, a second notification that the at least one additional instance of the serverless function has been destroyed.
2. The method of claim 1 , wherein the network component is a network switch.
3. The method of claim 2 , wherein the performance bottleneck is in a link connecting the network switch to at least one of the one or more hosts.
4. The method of claim 2 , wherein the performance bottleneck is at least one buffer saturation at the network switch.
5. The method of claim 2 , wherein the first message is sent to the network switch and at least one other network switch to which the one or more hosts are connected.
6. The method of claim 1 , wherein the first message identifies the host on which the at least one additional instance of the serverless function is being executed.
7. The method of claim 1 , wherein the first notification is generated in response to predicting the performance bottleneck, the performance bottleneck being predicted using one or more of statistic collected on the network component and a machine learning model trained using historical network performance data.
8. A device comprising:
memory configured to store computer-readable instructions therein; and
one or more processors configured to execute the computer-readable instructions to:
receive a first notification from a network component, the first notification indicating a performance bottleneck in association with one or more instances of a serverless function being executed at one or more hosts of a network;
initiate at least one additional instance of the serverless function in response to the performance bottleneck; and
send a first message to the network component identifying the at least one additional instance of the serverless function, the network component directing network traffic based on the first message;
determine whether the at least one additional instance of the serverless function is still needed;
in response to determining the at least one additional instance of the serverless function is not still needed, send a second message to the network component and host handling the at least one additional instance of the serverless function, the second message informing the network component and host the at least one additional instance of the serverless function is to be destroyed; and
receive a second notification that the at least one additional instance of the serverless function has been destroyed.
9. The device of claim 8 , wherein the network component is a network switch.
10. The device of claim 9 , wherein the performance bottleneck is in a link connecting the network switch to at least one of the one or more hosts.
11. The device of claim 9 , wherein the performance bottleneck is at least one buffer saturation at the network switch.
12. The device of claim 9 , wherein the first message is sent to the network switch and at least one other network switch to which the one or more hosts are connected.
13. The device of claim 8 , wherein the first message identifies a host on which the at least one additional instance of the serverless function is being executed.
14. The device of claim 8 , wherein the first notification is generated in response to predicting the performance bottleneck, the performance bottleneck being predicted using one or more of statistic collected on the network component and a machine learning model trained using historical network performance data.
15. A non-transitory computer-readable medium comprising computer-readable instructions stored therein, which when executed by at least one processor of a network orchestrator, cause the network orchestrator to:
receive a first notification from a network component, the first notification indicating a performance bottleneck in association with one or more instances of a serverless function being executed at one or more hosts of a network;
initiate at least one additional instance of the serverless function in response to the performance bottleneck;
send a first message to the network component identifying the at least one additional instance of the serverless function, the network component directing network traffic based on the first message;
determine whether the at least one additional instance of the serverless function is still needed;
in response to determining the at least one additional instance of the serverless function is not still needed, send a second message to the network component and host handling the at least one additional instance of the serverless function, the second message informing the network component and host the at least one additional instance of the serverless function is to be destroyed; and
receive a second notification that the at least one additional instance of the serverless function has been destroyed.
16. The non-transitory computer-readable medium of claim 15 , wherein the network component is a network switch.
17. The non-transitory computer-readable medium of claim 16 , wherein the performance bottleneck is in a link connecting the network switch to at least one of the one or more hosts.
18. The non-transitory computer-readable medium of claim 16 , wherein the performance bottleneck is at least one buffer saturation at the network switch.
19. The non-transitory computer-readable medium of claim 16 , wherein the first message is sent to the network switch and at least one other network switch to which the one or more hosts are connected.
20. The non-transitory computer-readable medium of claim 15 , wherein the first message identifies the host on which the at least one additional instance of the serverless function is being executed.