Automated downscaling of data stores
A process for facilitating downscaling of datastores (e.g., in a stateful system) is described herein. In embodiments, a set of metrics associated with a set of data stores of a stateful service is obtained. The set of metrics may indicate a utilization of each of the data stores of the set of data stores. Based on the set of metrics indicating underutilization associated with at least a portion of the set of data stores, a determination is made to initiate a downscaling of the set of data stores. Thereafter, a downscaler is deployed to perform downscaling operations to downscale the set of data stores. The downscaler communicates with a first data store to replicate data of the first data store onto a second data store. Based on identifying that the downscaler has completed the downscaling operations to downscale the set of data stores, the downscaler is terminated.
1. A computer-implemented method comprising:
obtaining a set of metrics associated with a set of data stores of a stateful service, the set of metrics indicating a utilization of each of the data stores of the set of data stores;
based on the set of metrics indicating underutilization associated with at least a portion of the set of data stores, determining, via a controller, to initiate a downscaling of the set of data stores;
in accordance with determining to initiate the downscaling of the set of data stores, deploying a downscaler, via a scaling of a pod separate from the controller, to perform downscaling operations to downscale the set of data stores, the downscaler communicating with a first data store to replicate data of the first data store onto a second data store of the set of data stores of the stateful service;
identifying that the downscaler has completed the downscaling operations to downscale the set of data stores, wherein identifying that the downscaler has completed the downscaling operations comprises identifying a lack of obtaining downscaling metrics within a predetermined period of time; and
based on the completion of the downscaling operations, terminating the downscaler.
2. The computer-implemented method of claim 1 , wherein the set of metrics comprise memory utilization and/or utilization rates.
3. The computer-implemented method of claim 1 , wherein the set of metrics indicate underutilization associated with at least a portion of the set of data stores when metrics associated with at least one data store fall below a utilization threshold.
4. The computer-implemented method of claim 1 further comprising identifying the first data store as a data store to terminate.
5. The computer-implemented method of claim 1 , wherein the downscaler comprises the pod in a container orchestration environment.
6. The computer-implemented method of claim 1 , wherein the downscaler is deployed by modifying a replica value associated with the downscaler from zero to one.
7. The computer-implemented method of claim 1 , wherein the downscaler is terminated by modifying a replica value associated with the downscaler from one to zero.
8. The computer-implemented method of claim 1 , wherein when the downscaling operations are being performed, a draining state is configured to indicate the first data store is to maintain a read-only state.
9. The computer-implemented method of claim 1 further comprising:
identifying completion of replicating data of the first data store onto a second data store; and
based on the completion of replicating data, removing the first data store.
10. The computer-implemented method of claim 1 further comprising:
identifying completion of replicating data of the first data store onto a second data store, wherein completion of replicating data of the first data store is identified based on communicating with the downscaler; and
based on the completion of replicating data, removing the first data store.
11. The computer-implemented method of claim 1 , wherein upon deployment of the downscaler, the downscaler communicates downscaling metrics to indicate performance of downscaling operations, and wherein upon completion of performing the downscaling operations, the downscaler terminates communication of downscaling metrics.
12. A system comprising:
a data store including computer-executable instructions; and
one or more processors configured to execute the computer-executable instructions, wherein execution of the computer-executable instructions causes the system to:
obtain a set of metrics associated with a set of data stores of a stateful service, the set of metrics indicating a utilization of each of the data stores of the set of data stores;
based on the set of metrics indicating underutilization associated with at least a portion of the set of data stores, determine, via a controller, to initiate a downscaling of the set of data stores;
in accordance with determining to initiate the downscaling of the set of data stores, deploy a downscaler, via a scaling of a pod separate from the controller, to perform downscaling operations to downscale the set of data stores, the downscaler communicating with a first data store to replicate data of the first data store onto a second data store of the set of data stores of the stateful service;
identify that the downscaler has completed the downscaling operations to downscale the set of data stores, wherein identifying that the downscaler has completed the downscaling operations comprises identifying a lack of obtaining downscaling metrics within a predetermined period of time; and
based on the completion of the downscaling operations, terminate the downscaler.
13. The system of claim 12 , wherein the set of metrics comprise memory utilization and/or utilization rates.
14. The system of claim 12 , wherein the set of metrics indicate underutilization associated with at least a portion of the set of data stores when metrics associated with at least one data store fall below a utilization threshold.
15. The system of claim 12 , wherein the downscaler comprises a pod in the container orchestration environment.
16. The system of claim 12 , wherein the downscaler is deployed by modifying a replica value associated with the downscaler from zero to one.
17. Non-transitory computer-readable media including computer-executable instructions that, when executed by a computing system, cause the computing system to:
obtain a set of metrics associated with a set of data stores of a stateful service, the set of metrics indicating a utilization of each of the data stores of the set of data stores;
based on the set of metrics indicating underutilization associated with at least a portion of the set of data stores, determine, via a controller, to initiate a downscaling of the set of data stores;
in accordance with determining to initiate the downscaling of the set of data stores, deploy a downscaler, via a scaling of a pod separate from the controller, to perform downscaling operations to downscale the set of data stores, the downscaler communicating with a first data store to replicate data of the first data store onto a second data store of the set of data stores of the stateful service;
identify that the downscaler has completed the downscaling operations to downscale the set of data stores, wherein identifying that the downscaler has completed the downscaling operations comprises identifying a lack of obtaining downscaling metrics within a predetermined period of time; and
based on the completion of the downscaling operations, terminate the downscaler.
18. The non-transitory computer-readable media of claim 17 , further comprising:
identifying completion of replicating data of the first data store onto a second data store, wherein completion of replicating data of the first data store is identified based on communicating with the downscaler; and
based on the completion of replicating data, removing the first data store.