A Hyper-V capability that hands an entire PCIe device, a graphics card or NVMe drive for example, directly to a single virtual machine.
Also called DDA.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Discrete Device Assignment in context, with comparison tables and the common traps.
Terms in this definition
- Hyper-V
The hypervisor built into Windows. The VMs for the Azure Extended Network appliance require the Hyper-V role, using nested virtualisation, and an external virtual switch for each NIC.
- Capability
Something that a person, organisation or system is able to do.
- PCIe
Bus standard used by NVMe drives, GPUs and similar cards. With Discrete Device Assignment, one VM gets an entire such device to itself.
- NVMe
Flash Optimized Azure Managed Redis keeps keys and hot values in memory while moving colder values, roughly 80 percent of capacity, onto this high-speed flash storage.
- VM
A computer implemented entirely in software, having virtual hardware and its own guest OS; ESX hosts run it and vCenter manages it.
Related terms
- IOMMU
Input-output memory management unit, a processor capability that maps how devices access memory. Discrete Device Assignment and GPU partitioning both require it.
- MMIO
Memory-mapped I/O space that a VM sets aside for devices. GPUs passed through with Discrete Device Assignment frequently need more than the starting allocation of 512 MB high and 128 MB low.