Dynamic recommendations taken over time for reservations of information technology resources

A system for modifying a recommendation of an optimal reservation number of computing resources, the system comprising:

• a server computer that includes:
• a network interface for communicating over one or more of a wired network or a wireless network with a client computer;
• a memory for storing a resource recommender software program;
• and one or more hardware processors that instantiate and execute the resource recommender software program that is programmed to perform actions, including:
• tracking historical usage of each of a plurality of the computing resources over time buckets that represent units of time, wherein the tracked historical usage includes:
• changes in a configuration of the computing resources, up time of the computing resources, down time of the computing resources, and loads placed on the computing resources;
• identifying, by machine learning algorithms temporary peaks in the usage of the computing resources using the tracked historical usage, to predict future computing resource usage over the time buckets, wherein the prediction include determinations of:
• when to release the computing resources, when to reallocate the computing resources, when to move an in-house resource to a cloud resource, and when to move the cloud resource back to the in-house resource;
• providing a trade-off percentage of a reservation for each of the computing resources based on a sum of an upfront cost and a reserved rate multiplied by a time of the reservation, the sum further divided by a mathematical product between an on-demand rate and the time of the reservation;
• providing, by using the trade-off percentage, a reservation period of a computing resource, of the plurality of computing resources, wherein the reservation period represents a time before the computing resource costs less to purchase as the reservation when its cost is amortized over the reservation period;
• providing, based on the time buckets further ranked from largest to smallest and the trade-off percentage, a trade-off point representing a number of the computing resources where it is cost effective to purchase reserved computing resources over using on-demand computing resources;
• and
• determining a recommendation for an optimum forecast model based on the ranked time buckets, the trade-off percentage, the trade-off point, the optimal reservation number of the computing resources over the reservation period, service level agreements, contracts, and the historical usage, wherein the reserved computing resources cover the computing resources during normal operation and the on-demand computing resources cover the computing resources above the optimal reservation number of the computing resources;
• and
• instantiating a user interface on a client computer to display the recommendation and also display selectable control elements;
• and
• in response to an input provided by a user, immediately and dynamically modifying assumptions of the optimum forecast model for the recommendation to generate a new recommendation that is displayed to the user, wherein the new recommendation is based on the modified assumptions of the optimum forecast model that provide a new estimate of corresponding savings or additional costs when fewer of the computing resources are provided than a previously recommended optimal reservation number of the computing resources over the reservation period from one or more different types of providers for one or more portions of the computing resources to improve an understanding of the new recommendation.

A non-transitory computer readable storage medium for modifying a recommendation of an optimal reservation number of computing resources, on which is recorded computer executable instructions that, when executed by a server computer that includes a memory, a network interface, and one or more processors, cause the one or more processors to instantiate and execute a resource recommender software program that performs actions, comprising:

• tracking historical usage of each of a plurality of the computing resources over time buckets that represent units of time, wherein the tracked historical usage includes:
• changes in a configuration of the computing resources, up time of the computing resources, down time of the computing resources, and loads placed on the computing resources;
• identifying, by machine learning algorithms temporary peaks in the usage of the computing resources using the tracked historical usage to predict future computing resource usage over the time buckets, wherein the prediction include determinations of:
• when to release the computing resources, when to reallocate the computing resources, when to move an in-house resource to a cloud resource, and when to move the cloud resource back to the in-house resource;
• providing, by the one or more processors, a trade-off percentage of a reservation for each of the computing resources based on a sum of an upfront cost and a reserved rate multiplied by a time of the reservation, the sum further divided by a mathematical product between an on-demand rate and the time of the reservation;
• providing, by using the trade-off percentage, a reservation period of a computing resource, of the plurality of computing resources, wherein the reservation period represents a time before the computing resource costs less to purchase as the reservation when its cost is amortized over the reservation period;
• providing, based on the time buckets further ranked from largest to smallest and the trade-off percentage, a trade-off point representing a number of the computing resources where it is cost effective to purchase reserved computing resources over using on-demand computing resources;
• and
• determining a recommendation for an optimum forecast model based on the ranked time buckets, the trade-off percentage, the trade-off point, the optimal reservation number of the computing resources over the reservation period, service level agreements, contracts, and the historical usage, wherein the reserved computing resources cover the computing resources during normal operation and the on-demand computing resources cover the computing resources above the optimal reservation number of the computing resources;
• and
• instantiating a user interface on a client computer to display the recommendation and also display selectable control elements;
• and
• in response to an input provided by a user, immediately and dynamically modifying assumptions of the optimum forecast model for the recommendation to generate a new recommendation that is displayed to the user, wherein the new recommendation is based on the modified assumptions of the optimum forecast model that provide a new estimate of corresponding savings or additional costs when fewer of the computing resources are provided than a previously recommended optimal reservation number of the computing resources over the reservation period from one or more different types of providers for one or more portions of the computing resources to improve an understanding of the new recommendation.