Nutanix AHV with NVIDIA Virtual GPU Solutions Ready to Meet the Demands of Any Workload Malcolm Crossley , AHV GPU Architect Tanuja Ingale , AHV Product Manager GTC, Mar 2019
Nutanix: The Enterprise Cloud Company Our Mission: Nutanix makes IT infrastructure invisible with an enterprise cloud platform that delivers the agility and economics of the public cloud, without sacrificing the security and control of on-premises infrastructure. 11,490+ customers Founded in 2009 Over 145+ countries 6 continents 4,380+ employees NTNX on Nasdaq 2
IT Complexity is Hurting Business Infrastructure Process People Time consuming Difficult to scale Little time for innovation and upgrade to provision Requires Large upfront Multiple points IT specialists CapEx of failure 3
The Need for Desktop Virtualization (VDI) Comprehensive Simplify support and Centralized security to Reduce cost and Increase employee desktop virtualization enable choice of BYO protect sensitive complexity of app and productivity with for any use case devices information desktop management anywhere access 4
VDI and GPU workload evolution High Density High Performance VDI, Client/Server & business apps Complex Design & Visualization • • Nutanix Comprehensive desktop virtualization High-end design, CAD, Rendering, Ray tracing (M&E, Manufacturing) • 100s of general knowledge & task workers per • Compute + Expansion module Support 10s of specialized users (Design) per Compute + Expansion module • ~2000 VDI users per X rack units w/o GPU • acceleration Support AI, Deep ML, Compute workloads for Improved VR perf • 200+ VDI sessions with accelerated virtualized • graphics HPC workloads, Monte Carlo analysis 5
How IT complexity translates to GPU environments GPU Infrastructure Process People Operational & workflow Need predictable scaling Requires complexity experts for GPU VDI CapEx Underutilization workflows/sizing Silos for graphics & compute Underutilization of True pre- Demand for GPU accelerated emptive capabilities of GPU workloads 6
Innovating in Three Fronts for VDI/GPU accelerated env Infra CapEx Technology Architecture Preservation Purchase based on Powerful data fabric Scale out, software- user needs & and control fabric defined architecture provide 100% capabilities utilization 7
Innovating in Three Fronts for VDI/GPU accelerated env Infra CapEx Technology Architecture Preservation Purchase based on Powerful data fabric Scale out, software- user needs & and control fabric defined architecture provide 100% capabilities utilization 8
VDI Pain points on legacy architecture • Commoditized by multicore Servers processors and dense memory • Managed by Hypervisor • Complex to manage • Costly to scale Storage Network • Performance bottleneck • Difficult to scale • Managed separately Centralized Storage (SAN/NAS) 9
Nutanix: Web-Scale Converged Architecture App App App App Virtualization Virtualization Server Server Storage Storage Storage Storage Storage Storage Controller Controller Controller Controller Controller Controller Integrated, scale-out compute and storage Built-in virtualization and management 10
Built for Virtualization Nutanix Controller VM (one per node) ✓ High performance ✓ Auto optimized ▪ Software-defined approach with Controller VM per node ▪ Pooled storage resources across the platform and scale as needed 11
The Result is Linearly Scaling Pay-as-you-grow VMs (Desktops) Number of Nodes • Scale incrementally one node at a time • Protect infrastructure investment by eliminating forklift upgrades • Scale storage capacity & performance linearly 12
Nutanix Platform Acropolis Prism Infrastructure / VM App Mobility Fabric Management Operational Distributed Storage Fabric Insights Capacity Planning AHV (built-in Virtualization) Comprehensive management solution radically Turnkey infrastructure platform that converges compute, storage and virtualization to simplifying datacenter operations run any application, at any scale 13
Nutanix AHV The hypervisor built for the Enterprise Cloud 14
AHV: Foundation of the Enterprise Cloud OS + + = Lean Enterprise Amazing Cloud Operating Virtualization Feature Set Management System NX Appliance OEM 3rd Party 15
AHV Powers HCI Automation Analytics Security • Prism Pro • • Distributed Scheduler Data Encryption • Performance • • Calm Blueprints Certifications Biz. Continuity Self-Healing Extendable • • HA Hot Spot Remediation • Memory • • Synch Rep (2019) Auto STIG Compliance • vCPU Auditability Management VDI/ GPU Acceleration • • Logging Prism Central • • Remote Syslog API and CLI • GPU, vGPU • Ecosystem PVS, MCS Performance • • Over 100 ISV solutions AHV Turbo • • Backup, SAP HANA IO Optimization 16
AHV: Citrix App and Desktop Virtualization Starting with Citrix XD/XA 7.9 … AHV is integrated with MCS Citrix Infrastructure End user desktops • XenApp & XenDesktop Nutanix • NetScaler VPX VM VM VM VM VM • ShareFile AHV is • XenApp, XenDesktop support VM VM VM VM VM • Cloud Connect InstantOn • PVS and MCS plug-ins • NVIDIA vGPU support • High perf. + data locality Nutanix AHV 17
Nutanix-NVIDIA Strategic Partnership • Nvidia: Industry leader in Visual Computing Technologies and GPU accelerators • AHV • First commercial kernel-based virtual machine (KVM) to support vGPU • Fully supports NVIDIA virtual GPU technology (GRID) • Quadro Virtual Data Center Workstation (vDWS) • nVIDIA GRID Virtual PC • GRID Virtual Applications 18
AHV: Crystal Clear Graphics vGPU Live vGPU GPU vGPU Multi- Migration HA vGPU VM (Coming Soon) (Coming Soon) Tesla M10. Tesla M60 Tesla P4 Tesla P40. Tesla v100 Tesla T4 19
AHV: Modes of GPU usage VM VM VM VM Guest OS Guest OS Guest OS Guest OS Apps Apps Apps Apps GPU driver GPU driver GPU driver GPU driver AHV Hypervisor NVIDIA GRID vGPU Manager Passthrough vGPU vGPU vGPU vGPU GRID GRID GPU GPU 20
Acropolis GPU resource management concepts NVIDIA Tesla M60 XenApp VM A Windows VM B Physical GPU Physical GPU 86:00.0 87:00.0 GPU config: GPU config: Vendor: Nvidia Vendor:Nvidia Vendor: Nvidia Vendor: Nvidia Type: PT-Graphics Type: PT-Compute Type: PT-Graphics Type: PT-Compute Device:M60 Device:M60 Device: M60 Device: M60 NVIDIA Tesla M10 Linux VM D Linux VM C Physical GPU Physical GPU 06:00.0 07:00.0 GPU config: GPU config: 2 x 2 x Vendor: Nvidia Vendor:Nvidia Vendor: Nvidia Vendor: Nvidia Type: PT-Graphics Type: PT-Graphics Type: PT-Graphics Type: PT-Graphics Device:M60 Device:M10 Device: M10 Device: M10 Not enough GPU resource Physical GPU Physical GPU 08:00.0 09:00.0 Vendor: Nvidia Vendor: Nvidia Type: PT-Graphics Type: PT-Graphics Device: M10 Device ID: M10 21
Acropolis GPU resource management concepts NVIDIA Tesla M60 XenApp VM D Physical GPU Physical GPU GPU config: 87:00.0 86:00.0 Vendor: Nvidia Type: PT-Graphics Xendesktop VM A Device: M60 Vendor: Nvidia Vendor: Nvidia GPU config: Type: PT-Graphics Type: PT-Compute Device: M60 Device: M60 Vendor: Nvidia Type: Virtual Device: M60-2Q Virtual GPU Virtual GPU Virtual GPU Virtual GPU 87:00.0 87:00.0 Virtual GPU 86:00.0 86:00.0 Virtual GPU 86:00.0 Vendor: Nvidia Vendor: Nvidia Xendesktop VM C 86:00.0 Vendor: Nvidia Vendor: Nvidia Type: Virtual Type: Virtual GPU config: Type: Virtual Vendor: Nvidia Type: Virtual Device: M60-1Q Device: M60-2Q Vendor: Nvidia Type: Virtual Device: M60-1Q Device: M60-1Q Index: 0 Index: 0 Vendor: Nvidia Type: Virtual Device: M60-1Q Index: 1 Index: 1 Type: Virtual Xendesktop VM B Device: M60-1Q Device ID: M60-1Q Index: 2 Index: 3 GPU config: Vendor: Nvidia Not enough GPU resource Type: Virtual Device: M60-2Q 22
AHV VM GPU resource configuration 23
Prism UI - Physical GPU overview 24
Prism UI - Physical GPU metrics 25
Prism UI - Virtual GPU metrics 26
Prism UI – Multi-vGPU per VM 27
Powerful REST API for all GPU resource information 28
Acropolis 1-click operations and GPU resources GPU GPU UVM VM VM AHV version 1 AHV version 2 AHV version 2 AHV version 1 GPU VM AHV version 1 AHV version 2 29
Innovating in Three Fronts for VDI/GPU accelerated env Infra CapEx Technology Architecture Preservation Purchase based on Powerful data fabric Scale out, software- user needs & and control fabric defined architecture provide 100% capabilities utilization 30
Nutanix Data & Control Fabric Solutions Software-defined Snapshots & Clones Relative Application Performance Offloads virtualization tier → higher ops performance 1.8 1.6 1.4 Array-based quick-clones for efficient provisioning 1.2 Seconds 1 Native VM-centric snapshots 0.8 0.6 0.4 0.2 Nutanix Shadow Clones 0 300 600 1200 1500 3000 Number of Virtual Desktops Distributed caching of vDisksand VM data read by multiple CVMs Consistent response time while incrementally scaling blocks ~50% reduction in boot time 31
Recommend
More recommend
