Package org.cloudbus.cloudsim.schedulers.vm

Interface VmScheduler

All Known Implementing Classes:
VmSchedulerAbstract, VmSchedulerSpaceShared, VmSchedulerTimeShared
public interface VmScheduler
An interface that represents the policy used by a Virtual Machine Monitor (VMM) to share processing power of a PM among VMs running in a host. Each host has to use its own instance of a VmScheduler that will so schedule the allocation of host's PEs for VMs running on it.

It also implements the Null Object Design Pattern in order to start avoiding NullPointerException when using the NULL object instead of attributing null to VmScheduler variables.

Since:
CloudSim Plus 1.0
Author:
Rodrigo N. Calheiros, Anton Beloglazov, Manoel Campos da Silva Filho

  • Field Details

    • LOGGER

      static final org.slf4j.Logger LOGGER

    • DEF_VM_MIGRATION_CPU_OVERHEAD

      static final double DEF_VM_MIGRATION_CPU_OVERHEAD
      The default percentage to define the CPU overhead of VM migration if one is not explicitly set.
      See Also:

    • NULL

      static final VmScheduler NULL
      An attribute that implements the Null Object Design Pattern for VmScheduler objects.

  • Method Details

    • allocatePesForVm

      boolean allocatePesForVm(Vm vm, MipsShare requestedMips)
      Requests the allocation of PEs for a VM.
      Parameters:
      vm - the vm to allocate PEs to
      requestedMips - the list of MIPS share to be allocated to a VM
      Returns:
      true if the PEs were allocated to the VM, false otherwise
      To do something:
      All implementing classes don't consider the situation when a Vm already has allocated MIPS and the method is called again. In this case, what is supposed to do? Increase the current allocation or change it? I think that the obvious action is to change the allocation, however, the implementations aren't working to deal with this situation. For that, they have to use some method such as Resource.isAmountAvailable(long) to first check if the difference from the current allocated mips and the requested one is available. Currently the implementations wrongly check if the total requested mips is available, while only the difference has to be checked. It has to be added some tests to check this issue.

    • allocatePesForVm

      boolean allocatePesForVm(Vm vm)
      Requests the allocation of PEs for a VM, according to the number of PEs and MIPS defined by VM attributes.
      Parameters:
      vm - the vm to allocate PEs to
      Returns:
      true if the PEs were allocated to the VM, false otherwise

    • deallocatePesFromVm

      void deallocatePesFromVm(Vm vm)
      Releases all PEs allocated to a VM. After that, the PEs may be used on demand by other VMs.
      Parameters:
      vm - the vm to deallocate PEs from

    • deallocatePesFromVm

      void deallocatePesFromVm(Vm vm, int pesToRemove)
      Releases a given number of PEs from a VM. After that, the PEs may be used on demand by other VMs.
      Parameters:
      vm - the vm to deallocate PEs from
      pesToRemove - number of PEs to deallocate

    • getAllocatedMips

      MipsShare getAllocatedMips(Vm vm)
      Gets the MIPS share of each host's Pe that is allocated to a given VM.
      Parameters:
      vm - the vm to get the MIPS share
      Returns:

    • getTotalAvailableMips

      double getTotalAvailableMips()
      Gets the total amount of MIPS that is currently free. If there are VMs migrating into the Host, their requested MIPS will already be allocated, reducing the total available MIPS.
      Returns:

    • getRequestedMips

      MipsShare getRequestedMips(Vm vm)
      Gets the List of MIPS requested by a VM.
      Parameters:
      vm - the VM to get the List of requested MIPS
      Returns:

    • isSuitableForVm

      boolean isSuitableForVm(Vm vm)
      Checks if the PM using this scheduler has enough MIPS capacity to host a given VM.
      Parameters:
      vm - the vm to check if there is enough available resource on the PM to host it
      Returns:
      true, if it is possible to allocate the the VM into the host; false otherwise
      See Also:

    • isSuitableForVm

      boolean isSuitableForVm(Vm vm, MipsShare requestedMips)
      Checks if a list of MIPS requested by a VM is allowed to be allocated or not. Depending on the VmScheduler implementation, the return value of this method may have different effects:
      • true: requested MIPS can be allocated, partial or totally;
      • false: requested MIPS cannot be allocated because there is no availability at all or there is just a partial amount of the requested MIPS available and the VmScheduler implementation doesn't allow allocating less than the VM is requesting. If less than the required MIPS is allocated to a VM, it will cause performance degradation. Such situation defines an over-subscription situation which just specific VmSchedulers accept.
      Parameters:
      vm - the Vm to check if there are enough MIPS to allocate to
      requestedMips - a list of MIPS requested by a VM
      Returns:
      true if the requested MIPS List is allowed to be allocated to the VM, false otherwise

    • getTotalAllocatedMipsForVm

      double getTotalAllocatedMipsForVm(Vm vm)
      Gets the actual total allocated MIPS for a VM along all its allocated PEs. If the VM is migrating into the Host, then just a fraction of the requested MIPS is actually allocated, representing the overhead of the migration process.

      The MIPS requested by the VM are just actually allocated after the migration is completed.

      Parameters:
      vm - the VM to get the total allocated MIPS
      Returns:
      See Also:

    • getMaxCpuUsagePercentDuringOutMigration

      double getMaxCpuUsagePercentDuringOutMigration()
      Gets the max percentage of CPU a VM migrating out of this Host can use. Since there may be an overhead associated to the migration process (if the CPU overhead for VM migration is greater than 0), during the migration, the amount of MIPS the VM can use is reduced due to this overhead.
      Returns:
      the max percentage of CPU usage during migration (in scale from [0 to 1], where 1 is 100%)

    • getVmMigrationCpuOverhead

      double getVmMigrationCpuOverhead()
      Defines the percentage of Host's CPU usage increase when a VM is migrating in or out of the Host. The value is in scale from 0 to 1 (where 1 is 100%).
      Returns:
      the Host's CPU migration overhead percentage.

    • getHost

      Host getHost()
      Gets the host that the VmScheduler get the list of PEs to allocate to VMs.
      Returns:

    • setHost

      VmScheduler setHost(Host host)
      Sets the host that the VmScheduler get the list of PEs to allocate to VMs. A host for the VmScheduler is set when the VmScheduler is set to a given host. Thus, the host is in charge to set itself to a VmScheduler.
      Parameters:
      host - the host to be set
      Returns:
      Throws:
      IllegalArgumentException - when the scheduler already is assigned to another Host, since each Host must have its own scheduler
      NullPointerException - when the host parameter is null