Safe and agile rollouts in a network-accessible server infrastructure using slices
Methods, systems, and apparatuses manage rolling out of updates in a network-accessible server infrastructure which operates a plurality of instances of a supporting service. The supporting service is comprised by a plurality of service portions. The instances of the supporting service each include of the service portions. The instances of the supporting service are partitioned into a plurality of slices. Each instance is partitioned to include one or more of the slices, and each slice of an instance includes one or more of the service portions. A software update is deployed to the instances by applying the software update to the slices in a sequence such that the software update is applied to a same slice in parallel across the instances containing that same slice before being applied to a next slice, and waiting a wait time before applying the software domain to a next slice in the sequencing.
1. A method of rolling out updates to a network-accessible server infrastructure which operates a plurality of instances of a supporting service, the supporting service comprised by a plurality of service portions, the instances of the supporting service each including the plurality of service portions, the method comprising:
receiving an indication of a partition of the instances of the supporting service into a plurality of slices, each instance of the supporting service partitioned to include one or more slices of the plurality of slices, each slice of an instance of the supporting service including one or more of the service portions of the instance of the supporting service; and
deploying a software update to the plurality of instances of the supporting service by
applying the software update to the plurality of slices in a sequence such that the software update is applied to a same slice in parallel across the instances of the supporting service containing that same slice before being applied to a next slice and
waiting a wait time after each applying of the software update to a slice of the plurality of slices before applying the software domain to a next slice of the plurality of slices in the sequence.
2. The method of claim 1 , wherein a first slice of the plurality of slices has complete configuration diversity coverage of the network-accessible server infrastructure.
3. The method of claim 1 , wherein a first slice of the plurality of slices has at least one of:
complete configuration diversity coverage of server hardware configurations included in the network-accessible server infrastructure, or
complete configuration diversity coverage of server software configurations included in the network-accessible server infrastructure.
4. The method of claim 1 , further comprising:
slicing the instances of the supporting service into the plurality of slices such that the slices increase in size in the sequence.
5. The method of claim 1 , wherein said waiting comprises:
decreasing the wait time for each iteration of said waiting after a first iteration of said waiting.
6. The method of claim 1 , wherein said waiting comprises:
during the wait time following the applying of the software update to a slice, monitoring the slice for failures caused by the software update; and
in response to the determination of a failure during said monitoring, terminating said deploying of the software update.
7. The method of claim 1 , wherein in a first instance of the instances of the supporting service, a first slice of the plurality of slices includes a first set of the service portions, and
in a second instance of the instances of the supporting service, the first slice of the plurality of slices includes a second set of the service portions that is different from the first set of the service portions.
8. A system, comprising:
at least one processor circuit; and
at least one memory that stores program code configured to be executed by the at least one processor circuit, the program code comprising:
a deployment orchestrator configured to roll out updates in a network-accessible server infrastructure which operates a plurality of instances of a supporting service, the supporting service comprised by a plurality of service portions, the instances of the supporting service each including the plurality of service portions, the deployment orchestrator configured to
receive an indication of a partition of the instances of the supporting service into a plurality of slices, each instance of the supporting service partitioned to include one or more slices of the plurality of slices, each slice of an instance of the supporting service including one or more of the service portions of the instance of the supporting service, and
deploy a software update to the plurality of instances of the supporting service, the deployment orchestrator including
an update sequencer configured to apply the software update to the slices in a sequence such that the software update is applied to a same slice in parallel across the instances of the supporting service containing that same slice before being applied to a next slice and
an update bake timer configured to enact a wait time after each applying of the software update to a slice of the plurality of slices by the update sequencer before enabling the update sequencer to apply the software domain to a next slice of the plurality of slices in the sequence.
9. The system of claim 8 , wherein a first slice of the plurality of slices has complete configuration diversity coverage of the network-accessible server infrastructure.
10. The system of claim 8 , wherein a first slice of the plurality of slices has at least one of:
complete configuration diversity coverage of server hardware configurations included in the network-accessible server infrastructure, or
complete configuration diversity coverage of server software configurations included in the network-accessible server infrastructure.
11. The system of claim 8 , wherein the deployment orchestrator further comprises:
a server slicer configured to slice the instances of the supporting service into the plurality of slices such that the slices increase in size in the sequence.
12. The system of claim 8 , wherein the update bake timer is further configured to decrease the wait time for each enacting of the wait time subsequent to a first enactment of the wait time.
13. The system of claim 8 , wherein the update bake timer is further configured to:
receive an indication of the failure caused by the software update; and
terminate the deploying of the software update if an indication of the failure caused by the software update is received.
14. The system of claim 8 , wherein in a first instance of the instances of the supporting service, a first slice of the plurality of slices includes a first set of the service portions, and
in a second instance of the instances of the supporting service, the first slice of the plurality of slices includes a second set of the service portions that is different from the first set of the service portions.
15. A computer-readable storage medium having program instructions recorded thereon that, when executed by at least one processing circuit, perform a method on a first computing device for rolling out updates to a network-accessible server infrastructure which operates a plurality of instances of a supporting service, the supporting service comprised by a plurality of service portions, the instances of the supporting service each including the plurality of service portions, the method comprising:
receiving an indication of a partition of the instances of the supporting service into a plurality of slices, each instance of the supporting service partitioned to include one or more slices of the plurality of slices, each slice of an instance of the supporting service including one or more of the service portions of the instance of the supporting service; and
deploying a software update to the plurality of instances of the supporting service by
applying the software update to the slices in a sequence such that the software update is applied to a same slice in parallel across the instances of the supporting service containing that same slice before being applied to a next slice and
waiting a wait time after each applying of the software update to a slice of the plurality of slices before applying the software domain to a next slice of the plurality of slices in the sequence.
16. The computer-readable storage medium of claim 15 , wherein a first slice of the plurality of slices has complete configuration diversity coverage of the network-accessible server infrastructure.
17. The computer-readable storage medium of claim 15 , further comprising:
slicing the instances of the supporting service into the plurality of slices such that the slices increase in size in the sequence.
18. The computer-readable storage medium of claim 15 , wherein said waiting comprises:
decreasing the wait time for each iteration of said waiting after a first iteration of said waiting.
19. The computer-readable storage medium of claim 15 , wherein said waiting comprises:
during the wait time following the applying of the software update to a slice, monitoring the slice for failures caused by the software update; and
in response to the determination of a failure during said monitoring, terminating said deploying of the software update.
20. The computer-readable storage medium of claim 15 , wherein in a first instance of the instances of the supporting service, a first slice of the plurality of slices includes a first set of the service portions, and
in a second instance of the instances of the supporting service, the first slice of the plurality of slices includes a second set of the service portions that is different from the first set of the service portions.