Cloud data center with reduced energy consumption
Technology for computing number of active servers needed over time in a cloud/compute cluster includes the following operations (not necessarily in the following order): (i) determining the capacity of each VCE provisioned on the cloud against the resource guaranteed to that VCE; (ii) forecasting the resource needs over time using historical requests for each VCE flavor; and (iii) using the forecasted resource needs to determine the required number of future servers at some future time. Some embodiments of the present invention use a formula that accounts for the interplay among various parameter values of the VCE flavors and also the mapping of the needs of VCEs of various flavors to the capabilities of physical resources.
1. A computer-implemented method (CIM) for use in a data center that uses VCEs to perform computing jobs, the CIM comprising:
receiving a plurality of VCE (virtual computing environment) flavor definitions, with each flavor definition including a VCE parameter value for at least one of the following VCE parameters: (i) number of VCPUs (virtual central processing units), (ii) amount of main memory, (iii) number of virtual connections, (iv) virtual network bandwidth, (v) storage I/O (input/output) bandwidth for remote storage, and/or (vi) number of GPUs (graphics processing units);
receiving a historical VCE request data set including a plurality of historical demands demand data set, with each historical demand data set including: (i) a previous demand on the data center to provide a VCE having a VCE flavor value corresponding to one of the VCE flavor definitions, and (ii) a time and date when the previous demand was made;
forecasting a VCE demand value for the data center at a future time based, at least in part, upon the historical VCE request data set;
determining a server demand value for the data center at a future time based, at least in part, upon the VCE demand value;
determining that a number of servers in a ready status pool of the data center is less than the server demand value; and
responsive to the determination that determining that a number of servers in the ready status pool of the data center is less than the server demand value, moving at least one server from an unready status pool to the ready status pool;
wherein the forecasting of the VCE demand value includes:
using classes of VCEs being requested or demanded for estimation, and
using priorities of VCEs being requested or demanded for estimation, and
using resource utilization measures for VCEs,
using constraints on resource oversubscription,
using constraints on resource utilization, and
using fragmentation in node resources.
2. The CIM of claim 1 further comprising:
transitioning servers in the unready status pool to low power modes.
3. The CIM of claim 2 further comprising:
communicating with data center power cap management (PCM) to communicate any changes to power characteristics of servers the unready status pool.
4. The CIM of claim 2 further comprising:
moving some idle servers in unready pool to service migration needs for servers that need to be vacated.
5. The CIM of claim 2 further comprising:
moving some idle servers in unready pool to service migration needs for customers who want to reserve servers for use on a customer's project.