Transcript OnTimeMeasure Capabilities for GENI Experimenters

“VMLab” Testbed for Desktop Virtualization Experiments
Prasad Calyam1,2,3, Alex Berryman 2,3, Albert Lai3, Jayshree Ramanathan3, Matthew Honigford4
Ohio Supercomputer Center1, OARnet2, The Ohio State University3, VMware4
Point-of-contact: [email protected]
October 2012
Project Overview
Goal:
Virtual Classroom Lab Pilot
VDPilot: Feasibility study of hosting virtual desktops for classroom labs within a federated university system in Ohio
Support desktop virtualization experiments for research and education user
communities
Major Efforts:
Subjective Testing Manuals
(a) Desktop virtualization sandboxes for system administrators and educators
(b) R&D activities relating to VDC resource allocation and thin-client
performance benchmarking
(c) Virtual desktops for classroom lab user trials involving faculty and
students
(d) Evaluation of the feasibility to deploy computationally intensive
interactive applications (e.g., remote volume visualization) in virtual
desktops
(e) Educational laboratory course curriculum development involving desktop
virtualization exercises
Testing Survey Form
VDBench Objective Testing
R&D Motivations
• Need for Human, Network and Application awareness in cloud platforms
to support large user workloads
GENI Experimentation with OnTimeMeasure, OpenFlow and VDI
GEC10 Demo
Unified Resource Broker
Measurement Engine
“A Cloud Service Provider can provision adequate CPU and memory resources to a VD
in the cloud, but if the thin-client protocol configuration does not account for network health
degradations and application context, the VD will be unusable for the user.”
“There is a need for frameworks and tools that can estimate how many concurrent VD
requests can be handled on a given set of system and network resources at a data center
such that resource utilization is maximized, and at worst, the minimum user QoE is
guaranteed as negotiated in service level agreements (SLAs)”
Measurement
Plane
Packet
Capture
Active
Measurement
Service Engine
Connection
Active
Broker
Directory
System
Provisioning
Congestion
Detection
Fault
Detection
Routing Engine
Marker Packet
Packet/Flow
Handler
Inspector
Resource
Optimization
OpenFlow
Controller
Control
Plane
Load
Balancing
Control
Plane
Secure
Channel
Secure
Channel
Data Center
Thin-clients
Thin-client
OpenFlow
Switches
OpenFlow
Switch
Virtual Desktop
Flow tables
Group Tables
RDP/PCoIP Client
User Applications
Data Plane
Security Token
RDP/PCoIP Server
File System
Resources @ http://vmlab.oar.net
• Website acts as a project website and testbed gateway
• IBM BladeCenter S Chassis that acts as a data center which can
concurrently support up to ~50 VDs
• 2 IBM HS22 Intel blade servers, 2 quad-core CPU’s each, 32 GB of RAM
each, and 4 NICs each
• VMware VDI, VMware View, VMware Horizon
GEC14 Demo
Hypervisor
0
1
2
3
Length
Group ID
Transport Protocol
OpCode
VMware View Connection Server MAC Address
Smart Thin-client MAC Address
VMware View Connection Server IP Address
Smart Thin-client IP Address
Server Port
MAC Header
IP Header
Client Port
TCP Header
Marker Packet Header
Data
Virtual Desktops for HPC Applications
• Computationally-intensive visualization applications in science-and-engineering are remotely accessible to users via thin-clients in
VMLab
Benefits: improves user convenience, enables shared resources, fosters user mobility, …
• Health Application: (funded by NIH) Remote Interactive Volume Visualization Infrastructure for Researchers (RIVVIR) being developed for
The Ohio State University Medical Center to support Small Animal Imaging Center Shared Resource (SAISR) users – VMLab is
helping in improving usability experiments
• Manufacturing Application: (funded by NIST) Moldex3D demo application is being used by Polymer Ohio to educate local
industry professionals on how to use HPC environments to discover more efficient injection molding techniques – VMLab is helping
in staging/feasibility experiments
Health Application
*NSF Award Number CNS-1205658: This material is based upon work supported by
the National Science Foundation under Grant No. CNS-1205658, VMware and Dell.
Any opinions, findings, and conclusions or recommendations expressed in this
material are those of the author(s) and do not necessarily reflect the views of