Smoothing termination of cloud resources
When tasks require compute/storage resources for their execution, provide these resources in a cloud, so the tasks can be executed on cloud resources. When one or more tasks are added, additional cloud resources are required to accommodate this increased workload. Initiating new cloud resources introduces latency into the process. Exponential smoothing may slow the rate at which unused cloud resources are terminated; however, this is processor intensive. With present invention it is not necessary to track the history of multiple data points from each time interval. Rather, it is only necessary to keep record of the number of tasks that were added to the workload during the first time interval. The controller only needs to keep a record of time intervals in which new tasks are added to the workload. In the intervening intervals, in which no new tasks were added, a simplified version of exponential smoothing can be applied.
1 . A controller having a processor, configured to smooth termination of cloud resources, the processor configured to:
determine a number of tasks (A t-n ) that were added to a workload during a first time interval (t-n);
identify a first quantity of cloud resources unused by the end of the first time interval (t-n);
estimate a second quantity, being a subset of the first quantity, of cloud resources to be required in a second time interval (t); and
terminate a third quantity of unused cloud resources in the second time interval (t), the third quantity of unused cloud resources equal to the first quantity of cloud resources less the second quantity of cloud resources;
wherein the second quantity of cloud resources to be required in the second time interval (t) is estimated from a number of tasks expected to be added to a workload (F t ) during the second time interval (t), the number of tasks expected to be added to the workload (F t ) during the second time interval (t) being estimated from the equation
F t =(1−α) n-1 ( F t-n +α( A t-n −F t-n ))
where:
F t-n is a number of tasks that were expected to be added to the workload during the first time interval (t-n);
n is the number of time intervals from the first time interval (t-n) to the second time interval (t); and
α is a smoothing factor having a value between zero and one.
2 . The controller of claim 1 , wherein each time interval is the same length as each other time interval.
3 . The controller of claim 1 , wherein the number of tasks expected to be added to the workload during the second time interval (t) is the integer part of F t from the equation
F t =(1−α) n-1 ( F t-n +α( A t-n −F t-n ))
4 . The controller of claim 1 , wherein a number of tasks that were expected to be added to the workload during an initial time interval, before which no previous estimations were made, may be zero.
5 . The controller of claim 1 , wherein a has a value of approximately 0.5.
6 . The controller of claim 1 , wherein the controller is further configured to:
determine whether a further number of tasks has been added to the workload since the first time interval;
in response to determining that no further tasks have been added to the workload since the first time interval, estimate the second quantity of cloud resources to be required in the second time interval as recited in claim 1 ; and
in response to determining that a second number (A t-m ) of tasks has been added to the workload since the first time interval (t):
identify a fourth quantity of cloud resources unused by the end of an intermediate time interval (t-m), the intermediate time interval (t-m) being the time interval in which the second number of tasks (A t-m ) was added to the workload; and
estimate the second quantity of cloud resources to be required in the second tie interval from the number of tasks expected to be added to a workload (F t ) during the second time interval (t), the number of tasks expected to be added to the workload (F t ) during the second time interval (t) being estimated from the equation
F t =(1−α) m-1 ( F t-m +α( A t-m −F t-m ))
where:
F t-m is a number of tasks that were expected to be added to the workload during the intermediate time interval (t-m);
m is the number of time intervals from the intermediate time interval (t-m) to the second time interval (t).
7 . A method of smoothing termination of cloud resources, the method comprising the steps of:
determining a number of tasks (A t-n ) that were added to a workload during a first time interval (t-n);
identifying a first quantity of cloud resources unused by the end of the first time interval (t-n);
estimating a second quantity, being a subset of the first quantity, of cloud resources to be required in a second time interval (t); and
terminating a third quantity of unused cloud resources in the second time interval (t), the third quantity of unused cloud resources equal to the first quantity of cloud resources less the second quantity of cloud resources;
wherein estimating the second quantity of cloud resources to be required in the second time interval (t) comprises estimating from a number of tasks expected to be added to a workload (F t ) during the second time interval (t), the number of tasks expected to be added to the workload (F t ) during the second time interval (t) being estimated from the equation
F t =(1−α) n-1 ( F t-n +α( A t-n −F t-n ))
where:
F t-n is a number of tasks that were expected to be added to the workload during the first time interval (t-n);
n is the number of time intervals from the first time interval (t-n) to the second time interval (t); and
α is a smoothing factor having a value between zero and one.
8 . A non-transitory computer-readable medium storing instructions for carrying out the method of claim 7 .