- Performance monitoring in KVM guests, BZ#645365
KVM can now virtualize a performance monitoring unit (vPMU) to allow virtual machines to use performance monitoring. Additionally it supports Intel's “architectural PMU” which can be live-migrated across different host CPU versions, using the
-cpu host flag.
With this feature, Red Hat virtualization customers are now able to utilize performance monitoring in KVM guests seamlessly. The virtual performance monitoring feature allows virtual machine users to identify sources of performance problems in their guests, using their preferred pre-existing profiling tools that work on the host as well as the guest. This is an addition to the existing ability to profile a KVM guest from the host.
This feature is a Technology Preview in Red Hat Enterprise Linux 6.3.
- Dynamic virtual CPU allocation
KVM in Red Hat Enterprise Linux 6.3 now supports dynamic virtual CPU allocation, also called vCPU hot plug, to dynamically manage capacity and react to unexpected load increases on their platforms during off-peak hours.
The virtual CPU hot-plugging feature gives system administrators the ability to dynamically adjust CPU resources in a guest. Because a guest no longer has to be taken offline to adjust the CPU resources, the availability of the guest is increased.
This feature is a Technology Preview in Red Hat Enterprise Linux 6.3. Currently, only the vCPU hot-add functionality works. The vCPU hot-unplug feature is not yet implemented.
- Virtio-SCSI capabilities
KVM Virtualization's storage stack has been improved with the addition of virtio-SCSI (a storage architecture for KVM based on SCSI) capabilities. Virtio-SCSI provides the ability to connect directly to SCSI LUNs and significantly improves scalability compared to virtio-blk. The advantage of virtio-SCSI is that it is capable of handling hundreds of devices compared to virtio-blk which can only handle 25 devices and exhausts PCI slots.
Virtio-SCSI is now capable of inheriting the feature set of the target device with the ability to:
attach a virtual hard drive or CD through the virtio-scsi controller,
pass-through a physical SCSI device from the host to the guest via the QEMU scsi-block device,
and allow the usage of hundreds of devices per guest; an improvement from the 32-device limit of virtio-blk.
This feature is a Technology Preview in Red Hat Enterprise Linux 6.3
- Support for in-guest S4/S3 states
KVM's power management features have been extended to include native support for S4 (suspend to disk) and S3 (suspend to RAM) states within the virtual machine, speeding up guest restoration from one of these low power states. In earlier implementations guests were saved or restored to/from a disk or memory that was external to the guest, which introduced latency.
Additionally, machines can be awakened from S3 with events from a remote keyboard through SPICE.
This feature is a Technology Preview and is disabled by default in Red Hat Enterprise Linux 6.3. To enable it, select the
/usr/share/seabios/bios-pm.bin file for the VM bios instead of the default
The native, in-guest S4 (suspend to disk) and S3 (suspend to RAM) power management features support the ability to perform suspend to disk and suspend to RAM functions in the guest (as opposed to the host), reducing the time needed to restore a guest by responding to simple keyboard gestures input. This also removes the need to maintain an external memory-state file. This capability is supported on Red Hat Enterprise Linux 6.3 guests and Windows guests running on any hypervisor capable of supporting S3 and S4.
- System monitoring via SNMP, BZ#642556
This feature provides KVM support for stable technology that is already used in data center with bare metal systems. SNMP is the standard for monitoring and is extremely well understood as well as computationally efficient. System monitoring via SNMP in Red Hat Enterprise Linux 6 allows the KVM hosts to send SNMP traps on events so that hypervisor events can be communicated to the user via standard SNMP protocol. This feature is provided through the addition of a new package: libvirt-snmp. This feature is introduced as a Technology Preview.
- Wire speed requirement in KVM network drivers
Virtualization and cloud products that run networking work loads need to run wire speeds. Up until Red Hat Enterprise Linux 6.1, the only way to reach wire speed on a 10 GB Ethernet NIC with a lower CPU utilization was to use PCI device assignment (passthrough), which limits other features like memory overcommit and guest migration
The macvtap/vhost zero-copy capabilities allow the user to use those features when high performance is required. This feature improves performance for any Red Hat Enterprise Linux 6.x guest in the VEPA use case. This feature is introduced as a Technology Preview.