Dynamic Topology Adaptation of Virtual Networks of Virtual Machines Ananth I. Sundararaj Ashish Gupta Peter A. Dinda Prescience Lab Department of Computer Science Northwestern University

2 Summary Dynamically adapt applications in virtual environments to available resources Demonstrate the feasibility of adaptation at the level of collection of VMs connected by VNET Show that its benefits can be significant for the case of BSP applications Studying the extent of applications for which our approach is effective

3 Outline Virtual machine grid computing Virtuoso system Networking challenges in Virtuoso Enter VNET VNET, VTTIFAdaptive virtual network Experiments Current Status Conclusions

4 Aim Grid Computing New Paradigm Traditional Paradigm Deliver arbitrary amounts of computational power to perform distributed and parallel computations Problem1: Grid Computing using virtual machines Problem2: Solution How to leverage them? Virtual Machines What are they? 6b 6a 5 4 3b 3a 2 1 Resource multiplexing using OS level mechanism Complexity from resource user’s perspective Complexity from resource owner’s perspective Virtual Machine Grid Computing

5 Virtual Machines Virtual machine monitors (VMMs) Raw machine is the abstraction VM represented by a single image VMware GSX Server

6 The Simplified Virtuoso Model Orders a raw machine User Specific hardware and performance Basic software installation available User’s LAN VM Virtual networking ties the machine back to user’s home network Virtuoso continuously monitors and adapts

7 User’s View in Virtuoso Model User User’s LAN VM

8 Outline Virtual machine grid computing Virtuoso system Networking challenges in Virtuoso Enter VNET VNET, VTTIFAdaptive virtual network Experiments Current Status Conclusions

9 User’s friendly LAN Foreign hostile LAN Virtual Machine Why VNET? A Scenario IP network User has just bought

10 User’s friendly LAN Foreign hostile LAN Virtual Machine VNET: A bridge with long wires Host Proxy X Why VNET? A Scenario VM traffic going out on foreign LAN IP network A machine is suddenly plugged into a foreign network. What happens? Does it get an IP address? Is it a routeable address? Does firewall let its traffic through? To any port?

11 User’s LAN Foreign LAN 1 Host 2 + VNET Proxy + VNET VNET startup topology IP network Host 3 + VNET Host 4 + VNET Host 1 + VNET Foreign LAN 3 Foreign LAN 4 Foreign LAN 2 VM 1 VM 4 VM 3 VM 2 TCP Connections

12 Host vmnet0 Ethernet Packet Tunneled over TCP/SSL Connection Ethernet Packet Captured by Interface in Promiscuous mode “Host Only” Network Ethernet Packet is Matched against the Forwarding Table on that VNET First linkSecond link (to proxy) Local traffic matrix inferred by VTTIF Periodically sent to the VNET on the Proxy VNET ethz VM “eth0” VNET ethy IP Network VM “eth0” vmnet0 A VNET Link

13 VTTIF Topology inference and traffic characterization for applications Ethernet-level traffic monitoring VNET daemons collectively aggregate a global traffic matrix for all VMs Application topology is recovered using normalization and pruning algorithms

14 VTTIF Operation Synced Parallel Traffic Monitoring Traffic Filtering and Matrix Generation Matrix Analysis and Topology Characterization

15 Dynamic Topology Inference 1. Fast updates Smoothed Traffic Matrix 2. Low Pass Filter Aggregation 3. Threshold change detection Topology change output VNET Daemons VTTIF parameters Update rate Smoothing interval Detection threshold

16 Reaction time of VTTIF

17 Outline Virtual machine grid computing Virtuoso system Networking challenges in Virtuoso Enter VNET VNET, VTTIFAdaptive virtual network Experiments Current Status Conclusions

18 Adaptation Virtuoso presents tremendous opportunities and challenges Adapt to available resources Challenges Network and host monitoring Monitor application Infer goals of application Adequacy of available mechanisms Challenges interrelated To determine subset of applications for which such adaptation succeeds We demonstrate that the subset is not empty

19 Experiments Focus on a specific instance –Application : Patterns, a synthetic benchmark –Monitoring : Application topology inferred by VTTIF –Aim : Minimize running time of patterns –Mechanism : Add links and corresponding forwarding rules to VNET topology Performance of BSP applications significantly enhanced by adapting VNET topology, guided by topology inferred by VTTIF

20 Foreign host LAN 1 User’s LAN Host 2 + VNET Proxy + VNET IP network Host 3 + VNET Host 4 + VNET Host 1 + VNET Foreign host LAN 3 Foreign host LAN 4 Foreign host LAN 2 VM 1 VM 4 VM 3 VM 2 Resilient Star Backbone Merged matrix as inferred by VTTIF Illustration of dynamic adaptation in Virtuoso Fast-path links amongst the VNETs hosting VMs

21 Evaluation Reaction time of VNET Benefits of adaptation (performance speedup) –Eight VMs on a single cluster, all-all topology –Eight VMs spread over two clusters over MAN, bus topology –Eight VMs spread over WAN, all-all topology

22 Reaction Time

23 Benefits of Adaptation Benefits accrued as a function of the number of fast-path links added Patterns has an all-all topology Eight VMs are used All VMs are hosted on the same cluster

24 Patterns has a bus topology Eight VMs are used VMs spread over two clusters over a MAN Benefits of Adaptation Benefits accrued as a function of the number of fast-path links added

25 Patterns has an all-all topology Eight VMs are used VMs are spread over WAN Benefits of Adaptation Benefits accrued as a function of the number of fast-path links added

26 Outline Virtual machine grid computing Virtuoso system Networking challenges in Virtuoso Enter VNET VNET, VTTIFAdaptive virtual network Experiments Current Status Conclusions

27 Current Status Applications: Transactional web ecommerce application Mechanisms: VM migration

28 Conclusions Demonstrated the feasibility of adaptation at the level of collection of VMs connected by VNET Showed that its benefits can be significant for the case of BSP applications Studying the extent of applications for which our approach is effective Moving ahead to use other adaptation mechanisms

29 For More Information –Prescience Lab (Northwestern University) –Virtuoso: Resource Management and Prediction for Distributed Computing using Virtual Machines VNET is publicly available from

30 Isn’t It Going to Be Too Slow? ApplicationResourceExecTime (10^3 s) Overhead SpecHPC Seismic (serial, medium) Physical16.4N/A VM, local % VM, Grid virtual FS % SpecHPC Climate (serial, medium) Physical9.31N/A VM, local % VM, Grid virtual FS % Experimental setup: physical: dual Pentium III 933MHz, 512MB memory, RedHat 7.1, 30GB disk; virtual: Vmware Workstation 3.0a, 128MB memory, 2GB virtual disk, RedHat 2.0 NFS-based grid virtual file system between UFL (client) and NWU (server) Small relative virtualization overhead; compute-intensive Relative overheads < 5%

31 Isn’t It Going To Be Too Slow? Synthetic benchmark: exponentially arrivals of compute bound tasks, background load provided by playback of traces from PSC Relative overheads < 10%

32 Isn’t It Going To Be Too Slow? Virtualized NICs have very similar bandwidth, slightly higher latencies –J. Sugerman, G. Venkitachalam, B-H Lim, “Virtualizing I/O Devices on VMware Workstation’s Hosted Virtual Machine Monitor”, USENIX 2001 Disk-intensive workloads (kernel build, web service): 30% slowdown –S. King, G. Dunlap, P. Chen, “OS support for Virtual Machines”, USENIX 2003 However: May not scale with faster NIC or disk