PROVISIONING OF VIRTUAL MACHINES WITH SECURITY REQUIREMENTS
Technologies are generally described to provision virtual machines with security requirements in datacenter. In some examples, a scheduler at a datacenter may receive a request to provision a virtual machine, where the virtual machine has an associated security requirement. Based on the security requirement, the scheduler may compute a maximum co-run probability of the virtual machine with at least one other virtual machine. The scheduler may then attempt to determine whether the virtual machine can be accommodated on an already-operational server while satisfying both the maximum co-run probability and a computing resource capacity associated with the virtual machine. If so, the virtual machine may be provisioned on the working server. Otherwise, the virtual machine may be provisioned on a new server if possible.
1 . A method to provision a virtual machine, the method comprising:
receiving a request to provision the virtual machine;
determining, based on a security requirement, a maximum co-run probability of another virtual machine with the virtual machine;
determining a computing resource capacity associated with the virtual machine; and
identifying a server on which the virtual machine is to be provisioned based on the maximum co-run probability and the computing resource capacity.
2 . The method of claim 1 , wherein determining the maximum co-run probability comprises determining the maximum co-run probability based on a number of virtual processing units available on the server.
3 . The method of claim 1 , wherein identifying the server comprises employing an equal scheduling scheme to identify the server.
4 . The method of claim 1 , wherein identifying the server comprises:
determining at least one server type that satisfies both the maximum co-run probability and the computing resource capacity; and
determining whether the server has a configuration similar to the determined at least one server type.
5 . The method of claim 1 , wherein identifying the server comprises launching a new server on which the virtual machine is to be provisioned.
6 . The method of claim 5 , wherein identifying the server further comprises evaluating at least one of an over-provisioning cost and a server launch cost.
7 . The method of claim 6 , wherein evaluating the at least one of the over-provisioning cost and the server launch cost comprises evaluating the over-provisioning cost and the server launch cost based on an estimated virtual machine start time and an estimated virtual machine time duration.
8 . The method of claim 7 , further comprising determining the estimated virtual machine start time and the estimated virtual machine time duration based on one or more of a linear regression estimation, a machine learning estimation, and a sliding window estimation.
9 . The method of claim 1 , further comprising provisioning the virtual machine on the identified server based on at least one of a bin-packing computation and an energy-aware heuristic computation.
10 . A virtual machine manager (VMM) configured to provision virtual machines, the VMM comprising:
a scheduler configured to:
receive a request to provision a virtual machine associated with a security requirement;
determine, based on the security requirement, a maximum co-run probability of another virtual machine with the virtual machine;
determine a computing resource capacity associated with the virtual machine; and
determine, based on the maximum co-run probability and the computing resource capacity, whether the virtual machine can be provisioned on a working server; and
a processor block configured to provision the virtual machine on the working server or cause the virtual machine to be provisioned on a new server.
11 . The VMM of claim 10 , wherein the scheduler is configured to determine the maximum co-run probability based on a number of virtual processing units available on the working server.
12 . The VMM of claim 10 , wherein the scheduler is configured to determine whether the virtual machine can be provisioned on the working server based on an equal scheduling scheme.
13 . The VMM of claim 10 , wherein the scheduler is further configured to:
determine at least one server type that satisfies both the maximum co-run probability and the computing resource capacity; and
determine whether the working server has a configuration similar to the determined at least one server type.
14 . The VMM of claim 10 , wherein the scheduler is further configured to evaluate at least one of an over-provisioning cost and a server launch cost to determine whether the virtual machine can be provisioned on the working server.
15 . The VMM of claim 14 , wherein the scheduler is configured to evaluate the over-provisioning cost and the server launch cost based on an estimated virtual machine start time and an estimated virtual machine time duration using one or more of a linear regression estimation, a machine learning estimation, and a sliding window estimation.
16 . The VMM of claim 10 , wherein the computing resource capacity includes one or more of a processor capacity, a processor core availability, a memory capacity, a bandwidth capacity, and a data storage capacity associated with the working server.
17 . The VMM of claim 10 , wherein the processor block is configured to receive the security requirement for one or more instances of the virtual machine.
18 . A cloud-based datacenter configured to provide cross-virtual-machine security, the datacenter comprising:
at least one working server configured to execute one or more virtual machines;
a scheduler configured to:
receive a request to provision a virtual machine associated with a security requirement;
determine, based on the security requirement, a maximum co-run probability of another virtual machine with the virtual machine;
determine a computing resource capacity associated with the virtual machine; and
determine, based on the maximum co-run probability and the computing resource capacity, whether the virtual machine can be provisioned on the at least one working server; and
a datacenter controller configured to one of:
provision the virtual machine on the at least one working server; and
start up a new server and provision the virtual machine on the new server.
19 . The datacenter of claim 18 , wherein the scheduler is configured to determine the maximum co-run probability based on a number of virtual processing units available on the at least one working server.
20 . The datacenter of claim 18 , wherein the scheduler is configured to determine whether the virtual machine can be provisioned on the at least one working server based on an equal scheduling scheme.
21 . The datacenter of claim 18 , wherein the scheduler is further configured to:
determine at least one server type that satisfies both the maximum co-run probability and the computing resource capacity; and
determine whether the at least one working server has a configuration similar to the determined at least one server type.
22 . The datacenter of claim 18 , wherein the scheduler is further configured to evaluate at least one of an over-provisioning cost and a server launch cost to determine whether the virtual machine can be provisioned on the at least one working server.
23 . The datacenter of claim 22 , wherein the scheduler is configured to evaluate the over-provisioning cost and the server launch cost based on an estimated virtual machine start time and an estimated virtual machine time duration.
24 . The datacenter of claim 23 , wherein the scheduler is further configured to determine the estimated virtual machine start time and the estimated virtual machine time duration based on one or more of a linear regression estimation, a machine learning estimation, and a sliding window estimation.
25 . The datacenter of claim 18 , wherein the datacenter controller is configured to receive the security requirement for one or more instances of the virtual machine.