1 Efficient Management of Data Center Resources for Massively Multiplayer Online Games V. Nae, A. Iosup, S. Podlipnig, R. Prodan, D. Epema, T. Fahringer,

Slides:



Advertisements
Similar presentations
Colyseus: A Distributed Architecture for Online Multiplayer Games
Advertisements

SLA-Oriented Resource Provisioning for Cloud Computing
Company name KUAS HPDS Using Remote Memory Paging for Handheld Devices in a Pervasive Computing Environment Arjuna Sathiaseelan.
Battle of Botcraft: Fighting Bots in Online Games with Human Observational Proofs Steven Gianvecchio, Zhenyu Wu, Mengjun Xie, and Haining Wang.
Implementation and Study of a “Term” based Role Playing Game using Client Server Paradigm. Vaithiyanathan Sundaram.
Network+ Guide to Networks, Fourth Edition
Proactive Prediction Models for Web Application Resource Provisioning in the Cloud _______________________________ Samuel A. Ajila & Bankole A. Akindele.
Nimesh Subramanian CMSC601.  Massively multiplayer online game (MMOG).  It is estimated that 55% of internet users play multiplayer online games. 
NUMA Tuning for Java Server Applications Mustafa M. Tikir.
A Service Platform for On-Line Games DebanJan Saha, Dambit Sahu, Anees Shaikh (IBM TJ Watson Research Center, NY) Presented by Gary Huang March 17, 2004.
Peer-to-Peer Support for Massively Multiplayer Games Bjorn Knutsson, Honghui Lu, Wei Xu, Bryan Hopkins Presented by Mohammed Alam (Shahed)
Improving Proxy Cache Performance: Analysis of Three Replacement Policies Dilley, J.; Arlitt, M. A journal paper of IEEE Internet Computing, Volume: 3.
1 IMPROVING RESPONSIVENESS BY LOCALITY IN DISTRIBUTED VIRTUAL ENVIRONMENTS Luca Genovali, Laura Ricci, Fabrizio Baiardi Lucca Institute for Advanced Studies.
1 Introduction to Load Balancing: l Definition of Distributed systems. Collection of independent loosely coupled computing resources. l Load Balancing.
Differentiated Multimedia Web Services Using Quality Aware Transcoding S. Chandra, C.Schlatter Ellis and A.Vahdat InfoCom 2000, IEEE Journal on Selected.
1 Measurement-based Characterization of a Collection of On-line Games Chris Chambers Wu-chang Feng Portland State University Sambit Sahu Debanjan Saha.
Network+ Guide to Networks, Fourth Edition Chapter 1 An Introduction to Networking.
Chapter 8: Network Operating Systems and Windows Server 2003-Based Networking Network+ Guide to Networks Third Edition.
.NET Mobile Application Development Introduction to Mobile and Distributed Applications.
Network Analysis of Counter-strike and Starcraft Mark Claypool, David LaPoint, Josh Winslow Worcester Polytechnic Institute Worcester, MA, USA
Niranjan Balasubramanian Aruna Balasubramanian Arun Venkataramani University of Massachusetts Amherst Energy Consumption in Mobile Phones: A Measurement.
Battle of Botcraft: Fighting Bots in Online Games withHuman Observational Proofs Steven Gianvecchio, Zhenyu Wu, Mengjun Xie, and Haining Wang The College.
1 Measurement-based Characterization of a Collection of On-line Games Chris Chambers Wu-chang Feng Portland State University Sambit Sahu Debanjan Saha.
Achieving Load Balance and Effective Caching in Clustered Web Servers Richard B. Bunt Derek L. Eager Gregory M. Oster Carey L. Williamson Department of.
STRATEGIES INVOLVED IN REMOTE COMPUTATION
Network+ Guide to Networks, Fourth Edition Chapter 1 An Introduction to Networking.
Storage Allocation in Prefetching Techniques of Web Caches D. Zeng, F. Wang, S. Ram Appeared in proceedings of ACM conference in Electronic commerce (EC’03)
University of Zagreb MMVE 2012 workshop1 Towards Reinterpretation of Interaction Complexity for Load Prediction in Cloud-based MMORPGs Mirko Sužnjević,
1 Scalable and transparent parallelization of multiplayer games Bogdan Simion MASc thesis Department of Electrical and Computer Engineering.
1 EuroPar 2009 – POGGI: Puzzle-Based Online Games on Grid Infrastructures POGGI: Puzzle-Based Online Games on Grid Infrastructures Alexandru Iosup Parallel.
Location-aware MapReduce in Virtual Cloud 2011 IEEE computer society International Conference on Parallel Processing Yifeng Geng1,2, Shimin Chen3, YongWei.
Peer-to-Peer Support for Massively Multiplayer Games Zone Federation of Game Servers : a Peer-to-Peer Approach to Scalable Multi-player Online Games [INFOCOM.
An Analysis of WoW Players’ Game Hours Matt Ross, Christian Ebinger, Anthony Morgan.
Copyright © 2011, Cost-Efficient Hosting and Load Balancing of Massively Multiplayer Online Games Nae, V.; Prodan, R.; Fahringer, T.; Grid Computing.
Challenges towards Elastic Power Management in Internet Data Center.
1 Wenguang WangRichard B. Bunt Department of Computer Science University of Saskatchewan November 14, 2000 Simulating DB2 Buffer Pool Management.
Dynamic Resource Monitoring and Allocation in a virtualized environment.
Energy Consumption in Mobile Phones: A Measurement Study and Implications for Network Applications REF:Balasubramanian, Niranjan, Aruna Balasubramanian,
1 ROIA 2009 – CAMEO: Continuous Analytics for Massively Multiplayer Online Games CAMEO: Continuous Analytics for Massively Multiplayer Online Games Alexandru.
Distributed Avatar Management for Second Life Matteo Varvello (Eurecom-Thomson) With Stefano Ferrari (Eurecom-Thomson), Ernst Biersack (Eurecom) Christophe.
Server Virtualization
1 Evaluation of Cooperative Web Caching with Web Polygraph Ping Du and Jaspal Subhlok Department of Computer Science University of Houston presented at.
1 MMORPG Servers. 2 MMORPGs Features Avatar Avatar Levels Levels RPG Elements RPG Elements Mission Mission Chatting Chatting Society & Community Society.
A P2P-Based Architecture for Secure Software Delivery Using Volunteer Assistance Purvi Shah, Jehan-François Pâris, Jeffrey Morgan and John Schettino IEEE.
Efficient AOI-Cast for Peer-to-Peer Networked Virtual Environments.
Pin-Yun Tarng / An Analysis of WoW Players’ Game Hours Network and Systems Laboratory nslab.ee.ntu.edu.tw IEEE/IFIP DSN 2008 Network and Systems Laboratory.
1 Adaptive Parallelism for Web Search Myeongjae Jeon Rice University In collaboration with Yuxiong He (MSR), Sameh Elnikety (MSR), Alan L. Cox (Rice),
Patch Scheduling for On-line Games Chris Chambers Wu-chang Feng Portland State University.
Author Utility-Based Scheduling for Bulk Data Transfers between Distributed Computing Facilities Xin Wang, Wei Tang, Raj Kettimuthu,
Ensieea Rizwani An energy-efficient management mechanism for large-scale server clusters By: Zhenghua Xue, Dong, Ma, Fan, Mei 1.
Predicting the Perceived Quality of a First Person Shooter Game The Team Fortress 2 T-Model David Dwyer Eric Finn Advisor: Mark Claypool 1.
1 TCS Confidential. 2 Objective : In this session we will be able to learn:  What is Cloud Computing?  Characteristics  Cloud Flavors  Cloud Deployment.
Capacity Planning in a Virtual Environment Chris Chesley, Sr. Systems Engineer
1 Evaluation of Cooperative Web Caching with Web Polygraph Ping Du and Jaspal Subhlok Department of Computer Science University of Houston presented at.
1 Performance Impact of Resource Provisioning on Workflows Gurmeet Singh, Carl Kesselman and Ewa Deelman Information Science Institute University of Southern.
Spark on Entropy : A Reliable & Efficient Scheduler for Low-latency Parallel Jobs in Heterogeneous Cloud Huankai Chen PhD Student at University of Kent.
Network Topologies for Scalable Multi-User Virtual Environments Lingrui Liang.
Dynamic Resource Allocation for Shared Data Centers Using Online Measurements By- Abhishek Chandra, Weibo Gong and Prashant Shenoy.
A Hierarchical Edge Cloud Architecture for Mobile Computing IEEE INFOCOM 2016 Liang Tong, Yong Li and Wei Gao University of Tennessee – Knoxville 1.
Introduction to Load Balancing:
Measurement-based Design
Memory Management for Scalable Web Data Servers
Vlad Nae, Radu Prodan, Thomas Fahringer Institute of Computer Science
CLUSTER COMPUTING.
Network+ Guide to Networks, Fourth Edition
Transparent Contribution of Memory
Measurement-based Characterization of a Collection of On-line Games
Resource Allocation for Distributed Streaming Applications
MagnaData: Scheduling Complex Workflows with Non-Functional Requirements in Datacenters Laurens Versluis Massivizing Computer research.
Transparent Contribution of Memory
Presentation transcript:

1 Efficient Management of Data Center Resources for Massively Multiplayer Online Games V. Nae, A. Iosup, S. Podlipnig, R. Prodan, D. Epema, T. Fahringer, SC’08 Shimin Chen Big Data Reading Group

2 Motivation Massively Multiplayer Online Games (MMOGs)  Popular in the past decade Existing approach:  Per game dedicated multi-server infrastructure E.g., World of Warcraft has > 10,000 servers  Over-provisioning for highly dynamic demands Daunting for new game providers to join the market!

3 This Paper Multiple games share many data centers  in-game monitoring Player interaction important  load prediction Neural network based  dynamic resource allocation  Geographical location of data centers with respect to user locations is taken into account

4 Outline Introduction A model for the MMOG ecosystem MMOG workload analysis Load prediction for MMOGs Resource provisioning and management Conclusion

5

6 MMOG Application Model Large-scale simulations of game worlds  Avatars: representation of players  NPCs or bots: non-player characters  Mobiles: other entities that can be interacted with  Decor: immutable entities Client-Server architecture:  Game providers maintain servers  Players run clients that connect to servers  Clients send play actions to servers  Servers compute game world state (positions of entities, interactions, etc.) then send responses to clients Smooth game experience is critical for success  Lack of responsiveness  people leaving the game  lose money

7 Characteristics Vary Greatly Interactions between players span a wide range  Very low: e.g., puzzle games  Low: e.g., RPG (role-playing game), a small group of players interact with a sparse environment  High: e.g., FPS (first-person shooter game), players are fighting against each other in a confined area The algorithms can there be O(n) – O(n 3 )  n: number of entities An common optimization is to only compute area of interest of each avatar

8 Data Center Model Many data centers scattered around the world Four types of resources:  CPU time, memory, ExtNetin, ExtNetout  (disk storage is not important?) Game providers submit resource requests Data centers allocate resources Each data center may enforce a particular size granularity for allocations (called space-time policy)  Requests are rounded up to this

9 Ecosystem Multiple data centers  Each may host multiple games Multiple game providers  Each may provide multiple games Each game may run on multiple data centers Resource allocation goals:  Allocated resources must match or larger than required  Locate resources closest to users  Select as finer-grain recourses with shorter reservation times as possible

10 Outline Introduction A model for the MMOG ecosystem MMOG workload analysis Load prediction for MMOGs Resource provisioning and management Conclusion

11 RuneScape Traces Ranked #2 by number of players in US and EU  5 million active players, 8 million open accounts  Combine elements of RPG and FPS  Game load cannot be trivially computed Trace: Aug 2007 – July 2008  Collected from the official RuneScape web page  Number of players over time for each server group  Record per two minutes

12 Number of concurrent players change greatly  over-provisioning

13 Region 0 (Europe) Loads 40 server groups, 2-week trace, sample / 2 minutes Strong diurnal pattern, but no weekend effects

14 Influence of Player Interaction on Load Tcpdump for 8 game sessions at clients

15 Inter-Arrival Time Large differences

16 Outline Introduction A model for the MMOG ecosystem MMOG workload analysis Load prediction for MMOGs Resource provisioning and management Conclusion

17 Two Options Explanatory models:  Tightly coupled with applications and platform  Difficult to obtain and update for dynamic complex systems Time series prediction  Neural network (another paper from the group)  Task: predicting entity counts of each sub-zone

18 Methodology MMOG Emulator:  Modeling different type of players  (No validation?) Test algorithms on the trace from the emulator

19 Prediction Results No sure if the time vs. prediction tradeoff favors the approach here?

20 Outline Introduction A model for the MMOG ecosystem MMOG workload analysis Load prediction for MMOGs Resource provisioning and management Conclusion

21 Experimental Setup Evaluation space:  Resource allocation mechanisms: static, dynamic  Prediction algorithms  Player interaction/model updates complexity  Hosting policies  Latency tolerance  Number of MMOGs Use RuneScape trace to model number of players and their distribution

22 Data Centers Emulation on real machines? Simulation? Not sure

23 Impact of Prediction Performance

24 Static vs. Dynamic Allocation Allocation granularity 6 hours

25 Impact of Player Interaction

26 Multiple MMOGs A: O(nlog(n)); B: O(n 2 ); C: O(n 2 log(n)) Clearly, the efficiency of the provisioning system is determined by the biggest consumer!

27 Conclusion Multiple MMOGs, multiple data centers Workload analysis Applying workload prediction to provisioning Emulation/simulation study  Not very convincing Ongoing work: