4A0-M05 practice question 1 of 15

Which of the following statements about SR-IOV is FALSE?

Full set: 56 questions
Answer options
A: VMs using OVS and SR-IOV can be instantiated on the same compute node.
B: SR-IOV offers comparable performance to non-visualized hardware.
C: The VM must be attached to the PF to achieve highest performance. Correct answer
D: Using SR-IOV restricts VM mobility.
Correct Answer:
C. The VM must be attached to the PF to achieve highest performance.
Explanation

The statement that a Virtual Machine (VM) must be attached to the Physical Function (PF) to achieve the highest performance is false. In an SR-IOV architecture, the PF is the full-featured PCIe device managed by the hypervisor or host operating system. The PF is capable of creating multiple lightweight Virtual Functions (VFs). For high-performance networking, VMs are assigned and directly attached to these VFs, not the PF. This direct assignment to a VF allows the VM to bypass the hypervisor's virtual switch, which is the mechanism that provides near-bare-metal performance. Attaching a VM to the entire PF would be PCI passthrough, a different technology.

Why Incorrect

A. This statement is true. A single compute node can concurrently host VMs using SR-IOV for performance-critical workloads and VMs using OVS for workloads requiring more network flexibility.
B. This statement is true. By bypassing the hypervisor's software switching layer and providing direct hardware access via VFs, SR-IOV significantly reduces latency and CPU overhead, achieving network performance comparable to bare-metal systems.
D. This statement is true. SR-IOV creates a direct binding between a VM and the physical NIC hardware on a specific host. This tight coupling complicates or prevents seamless live migration, thus restricting VM mobility.

References

1. Nokia White Paper, "The role of SR-IOV in a carrier-grade NFV platform" (2016). Section "SR-IOV Technology Overview", Page 3, states: "SR-IOV allows a physical PCIe device to present itself as multiple independent virtual devices, or Virtual Functions (VFs). The hypervisor assigns one or more VFs to a VM... The PF is managed by the hypervisor and is used to configure and manage the VFs." This confirms that VMs attach to VFs, not the PF, making statement C false. The paper also notes on Page 4, "The main drawback of SR-IOV is that it breaks the hardware abstraction layer... This makes it difficult to support features such as live migration of VMs," confirming statement D is true.

2. OpenStack Documentation, Networking Guide. In the section "SR-IOV for networking", it is detailed that "The SR-IOV architecture consists of the Physical Function (PF) and Virtual Functions (VFs)... The PF is a full-featured PCIe device, and the VFs are lightweight PCIe devices. VFs are assigned to guest instances [VMs]." This supports the reasoning that statement C is false. The guide also explicitly states, "Live migration is not supported for instances with SR-IOV ports," which validates that statement D is true.

3. Intel Corporation, "Ethernet Controller SR-IOV Primer" (Technical Brief). Page 2, "SR-IOV Architecture", explains: "The PF is a full-featured NIC that is discoverable, manageable, and configurable by the hypervisor... The VFs are lightweight NICs for direct assignment to VMs." This clearly refutes statement C. The document also highlights the performance benefits, supporting statement B.

4. Red Hat Enterprise Linux 8 Documentation, "Configuring and managing virtualization". Chapter 11, "Managing device passthrough", Section 11.3, "SR-IOV", clarifies: "The hypervisor can assign one or more VFs to a VM, and the VM accesses the device directly without mediation from the hypervisor." This confirms the mechanism for high performance and contradicts statement C. It also notes the limitation: "Using SR-IOV prevents the use of certain virtualization features, such as live migration." This confirms statement D is true.

Scroll to Top