Systems and methods for managing quality of service
The disclosed computer-implemented method for managing quality of service may include (i) installing, at a network protocol layer, a latency monitoring plug-in that monitors latency at the network protocol layer as distinct from a storage layer beneath the network protocol layer, (ii) providing, by the latency monitoring plug-in, feedback reporting the latency at the network protocol layer to a quality of service engine that monitors characteristics of the storage layer such that the quality of service engine learns about the latency at the network protocol layer, and (iii) adjusting, by the quality of service engine, a calculation of system capacity, based on both the feedback from the latency monitoring plug-in reporting the latency at the network protocol layer and the monitored characteristics of the storage layer to enable an application to adhere to a service level agreement. Various other methods, systems, and computer-readable media are also disclosed.
1. A computer-implemented method for managing quality of service, at least a portion of the method being performed by a computing device comprising at least one processor, the method comprising:
installing, at a network protocol layer, a latency monitoring plug-in that monitors latency at the network protocol layer as distinct from a storage layer beneath the network protocol layer;
providing, by the latency monitoring plug-in, feedback reporting the latency at the network protocol layer to a quality of service engine that monitors characteristics of the storage layer beneath the network protocol layer such that the quality of service engine learns about the latency at the network protocol layer; and
adjusting, by the quality of service engine, a calculation of system capacity, based on both the feedback from the latency monitoring plug-in reporting the latency at the network protocol layer and the monitored characteristics of the storage layer beneath the network protocol layer to enable an application to adhere to a service level agreement despite the latency at the network protocol layer;
wherein:
the feedback reports an average latency for a multitude of workloads to the quality of service engine;
adjusting the calculation further comprises calculating a network capacity based on the latency at the network protocol layer as latency perceived by the application and calculating a storage capacity based on the monitored characteristics of the storage layer; and
the method further comprises selecting the minimum between the network capacity and the storage capacity as the system capacity to enable the application to adhere to the service level agreement despite the latency at the network protocol layer.
2. The computer-implemented method of claim 1 , wherein the network protocol layer corresponds to the QUICK EMULATOR hypervisor.
3. The computer-implemented method of claim 1 , wherein the network protocol layer corresponds to the NETWORK BLOCK DEVICE.
4. The computer-implemented method of claim 1 , wherein the network protocol layer corresponds to the INTERNET SMALL COMPUTER SYSTEMS INTERFACE.
5. The computer-implemented method of claim 1 , wherein the latency at the network protocol layer is induced by an I/O path at the network protocol layer.
6. The computer-implemented method of claim 1 , wherein the network protocol layer corresponds to the NETWORK FILE SYSTEM.
7. The computer-implemented method of claim 1 , wherein the application corresponds to a guest virtual machine of a hypervisor.
8. The computer-implemented method of claim 7 , wherein the storage layer corresponds to storage shared by multiple guest virtual machines of the hypervisor.
9. The computer-implemented method of claim 7 , wherein at least one of:
an administrator specifies a latency tolerance for the guest virtual machine; and
the quality of service engine learns the latency tolerance for the guest virtual machine.
10. The computer-implemented method of claim 1 , wherein:
a previous inability of the quality of service engine to learn about the latency at the network protocol layer creates a potential noisy neighbor problem, according to a statistical latency analysis, when a neighbor consumes a network resource beyond a threshold; and
providing, by the latency monitoring plug-in, feedback reporting the latency at the network protocol layer to the quality of service engine such that the quality of service engine learns about the latency at the network protocol layer effectively solves the potential noisy neighbor problem.
11. A system for managing quality of service, the system comprising:
an installation module, stored in a memory, that installs, at a network protocol layer, a latency monitoring plug-in that monitors latency at the network protocol layer as distinct from a storage layer beneath the network protocol layer;
a provisioning module, stored in the memory, that provides, through the latency monitoring plug-in, feedback reporting the latency at the network protocol layer to a quality of service engine that monitors characteristics of the storage layer beneath the network protocol layer such that the quality of service engine learns about the latency at the network protocol layer;
an adjusting module, stored in the memory, that adjusts, as part of the quality of service engine, a calculation of system capacity, based on both the feedback from the latency monitoring plug-in reporting the latency at the network protocol layer and the monitored characteristics of the storage layer beneath the network protocol layer to enable an application to adhere to a service level agreement despite the latency at the network protocol layer; and
at least one physical processor configured to execute the installation module, the provisioning module, and the adjusting module;
wherein:
the feedback reports an average latency for a multitude of workloads to the quality of service engine;
the adjusting module adjusts the calculation at least in part by calculating a network capacity based on the latency at the network protocol layer as latency perceived by the application and calculating a storage capacity based on the monitored characteristics of the storage layer; and
the adjusting module selects the minimum between the network capacity and the storage capacity as the system capacity to enable the application to adhere to the service level agreement despite the latency at the network protocol layer.
12. The system of claim 11 , wherein the network protocol layer corresponds to the QUICK EMULATOR hypervisor.
13. The system of claim 11 , wherein the network protocol layer corresponds to the NETWORK BLOCK DEVICE.
14. The system of claim 11 , wherein the network protocol layer corresponds to the INTERNET SMALL COMPUTER SYSTEMS INTERFACE.
15. The system of claim 11 , wherein the latency at the network protocol layer is induced by an I/O path at the network protocol layer.
16. The system of claim 11 , wherein the network protocol layer corresponds to the NETWORK FILE SYSTEM.
17. The system of claim 11 , wherein the application corresponds to a guest virtual machine of a hypervisor.
18. The system of claim 17 , wherein the storage layer corresponds to storage shared by multiple guest virtual machines of the hypervisor.
19. The system of claim 17 , wherein at least one of:
an administrator specifies latency tolerance for the guest virtual machine; and
the quality of service engine learns the latency tolerance for the guest virtual machine.
20. A non-transitory computer-readable medium comprising one or more computer-readable instructions that, when executed by at least one processor of a computing device, cause the computing device to:
install, at a network protocol layer, a latency monitoring plug-in that monitors latency at the network protocol layer as distinct from a storage layer beneath the network protocol layer;
provide, by the latency monitoring plug-in, feedback reporting the latency at the network protocol layer to a quality of service engine that monitors characteristics of the storage layer beneath the network protocol layer such that the quality of service engine learns about the latency at the network protocol layer; and
adjust, by the quality of service engine, a calculation of system capacity, based on both the feedback from the latency monitoring plug-in reporting the latency at the network protocol layer and the monitored characteristics of the storage layer beneath the network protocol layer to enable an application to adhere to a service level agreement despite the latency at the network protocol layer;
wherein:
the feedback reports an average latency for a multitude of workloads to the quality of service engine;
adjusting the calculation further comprises calculating a network capacity based on the latency at the network protocol layer as latency perceived by the application and calculating a storage capacity based on the monitored characteristics of the storage layer; and
the instructions further cause the computing device to select the minimum between the network capacity and the storage capacity as the system capacity to enable the application to adhere to the service level agreement despite the latency at the network protocol layer.