[2017] virtio: vhost Data Path Acceleration towards NFV Cloud by Cunming Liang
VIRTIO as a para-virtualized device decouples VMs and physical devices. Friendly live-migration support makes it well recognized by the cloud networking. However, the north-south I/O throughput of VIRTIO doesn’t compare with SR-IOV.
The technical talk gives a practical proposal to address this by introducing a framework for vhost data path offloading. It essentially allows using HW device to offload the vring data path on the vhost side, without requiring the HW device being pass-thru to the guest. As a result, it achieves SR-IOV like performance with compatible live-migration support which makes it possible to upgrade a stock VM using virtio to a new HW accelerated platform transparently.
Steve is going to explain the mechanism under the hood, go through the detail proposal, introduce the benefits, and share the WIP patch work and figure out the minimum HW prerequisite for doing this.
---
Cunming Liang
Intel
Platform Solution Architect
Steve is Intel platform solution architecture for DPDK. He spends years on packet I/O performance optimization (e.g. vector PMD), I/O virtualization and core efficiency for DPDK. He used to work on the estimation/tuning/optimization of user space stack for workloads (e.g. L7 load balance). He addressed the virtualization/container story for DPDK, as one of the major input to define DPDK road map for that piece. The recent focus area is packet processing w/ HW acceleration or on the heterogeneous platform. He delivered a few presentation on DPDK China Summit, DPDK US Summit, OPNFV DPDK/FD.IO Mini Summit.
Видео [2017] virtio: vhost Data Path Acceleration towards NFV Cloud by Cunming Liang канала KVM Forum
The technical talk gives a practical proposal to address this by introducing a framework for vhost data path offloading. It essentially allows using HW device to offload the vring data path on the vhost side, without requiring the HW device being pass-thru to the guest. As a result, it achieves SR-IOV like performance with compatible live-migration support which makes it possible to upgrade a stock VM using virtio to a new HW accelerated platform transparently.
Steve is going to explain the mechanism under the hood, go through the detail proposal, introduce the benefits, and share the WIP patch work and figure out the minimum HW prerequisite for doing this.
---
Cunming Liang
Intel
Platform Solution Architect
Steve is Intel platform solution architecture for DPDK. He spends years on packet I/O performance optimization (e.g. vector PMD), I/O virtualization and core efficiency for DPDK. He used to work on the estimation/tuning/optimization of user space stack for workloads (e.g. L7 load balance). He addressed the virtualization/container story for DPDK, as one of the major input to define DPDK road map for that piece. The recent focus area is packet processing w/ HW acceleration or on the heterogeneous platform. He delivered a few presentation on DPDK China Summit, DPDK US Summit, OPNFV DPDK/FD.IO Mini Summit.
Видео [2017] virtio: vhost Data Path Acceleration towards NFV Cloud by Cunming Liang канала KVM Forum
Показать
Комментарии отсутствуют
Информация о видео
Другие видео канала
[2017] Fast Write-Protect and Fast Dirtylog-Bitmap Sync Up - Guangrong Xiao[2019] Firecracker: Lessons from the Trenches by Andreea Florescu and Alexandra Iordache[2014] KVM on MIPS by James Hogan[2017] Qemu Backup by Vladimir Sementsov-Ogievskiy & Maxim Nestratov[2017] vIOMMU/ARM: Full Emulation and virtio-iommu Approaches by Eric Auger[2020] KVM Dirty Ring - A New Approach to Logging by Peter Xu[2014] Towards Multithreaded Device Emulation in QEMU by Stefan Hajnoczi[2017] Running Virtual Machines on Kubernetes with libvirt & KVM by Fabian Deutsch & Roman Mohr[2014] Thanks for Live Snapshots, Where's Live Merge? by Adam Litke[2015] qcow2: why (not)? by Max Reitz and Kevin WolfKVM Forum 2013: oVirt for PowerPC - Leonardo Bianconi, Instituto de Pesquisas EldoradoKVM Forum 2013: RDMA Live Migration and RDMA Fault Tolerance by Michael Hines"Hybrid" Nesting: KVM on Hyper-V by Vitaly Kuznetsov & Tianyu Lan[2020] QEMU Snaphosts Are So Slow. Really? by Denis Lunev[2016] Virtualization Developer PanelVirtualizing Arm TrustZone on KVM by Chun Yen Lin & Shih-Wei Li[2017] Adding Device Direct Assignment Support for a New Bus Infrastructure on ARM64 by Mike Caraman[2020] Evaluate Implementation Options of KVM-based Type1 (or 1.5) Hypervisor by Jun NakajimaContributor Q&A PanelKVM: Arm Confidential Compute Architecture SupportSPDK vhost Target: A Practical Solution to Accelerate Storage I/Os by Ziye Yang & Changpeng Liu