Distributed Load Balancing for Key-Value Storage Systems Imranul Hoque Michael Spreitzer Malgorzata Steinder.

Slides:



Advertisements
Similar presentations
Ion Stoica, Robert Morris, David Karger, M. Frans Kaashoek, Hari Balakrishnan MIT and Berkeley presented by Daniel Figueiredo Chord: A Scalable Peer-to-peer.
Advertisements

Scalable Content-Addressable Network Lintao Liu
Fast Algorithms For Hierarchical Range Histogram Constructions
Clayton Sullivan PEER-TO-PEER NETWORKS. INTRODUCTION What is a Peer-To-Peer Network A Peer Application Overlay Network Network Architecture and System.
Computer Science Dr. Peng NingCSC 774 Adv. Net. Security1 CSC 774 Advanced Network Security Topic 7.3 Secure and Resilient Location Discovery in Wireless.
Dynamo: Amazon's Highly Available Key-value Store Distributed Storage Systems CS presented by: Hussam Abu-Libdeh.
Adaptive Storage Management for Modern Data Centers Imranul Hoque 1.
Small-world Overlay P2P Network
Kuang-Hao Liu et al Presented by Xin Che 11/18/09.
Overlay Networks + Internet routing has exhibited scalability - Internet routing is inefficient -Difficult to add intelligence to Internet Solution: Overlay.
Web Caching Schemes1 A Survey of Web Caching Schemes for the Internet Jia Wang.
A Comparison of Layering and Stream Replication Video Multicast Schemes Taehyun Kim and Mostafa H. Ammar.
Peer-to-Peer Based Multimedia Distribution Service Zhe Xiang, Qian Zhang, Wenwu Zhu, Zhensheng Zhang IEEE Transactions on Multimedia, Vol. 6, No. 2, April.
Efficient, Proximity-Aware Load Balancing for DHT-Based P2P Systems Yingwu Zhu, Yiming Hu Appeared on IEEE Trans. on Parallel and Distributed Systems,
Mercury: Scalable Routing for Range Queries Ashwin R. Bharambe Carnegie Mellon University With Mukesh Agrawal, Srinivasan Seshan.
Prefix Caching assisted Periodic Broadcast for Streaming Popular Videos Yang Guo, Subhabrata Sen, and Don Towsley.
A Scalable Content-Addressable Network Authors: S. Ratnasamy, P. Francis, M. Handley, R. Karp, S. Shenker University of California, Berkeley Presenter:
Efficient replica maintenance for distributed storage systems Byung-Gon Chun, Frank Dabek, Andreas Haeberlen, Emil Sit, Hakim Weatherspoon, M. Frans Kaashoek,
Distributed Lookup Systems
Scheduling with Optimized Communication for Time-Triggered Embedded Systems Slide 1 Scheduling with Optimized Communication for Time-Triggered Embedded.
A Row-Permutated Data Reorganization Algorithm for Growing Server-less VoD Systems Presented by Ho Tsz Kin.
1 Introduction to Load Balancing: l Definition of Distributed systems. Collection of independent loosely coupled computing resources. l Load Balancing.
1 An Empirical Study on Large-Scale Content-Based Image Retrieval Group Meeting Presented by Wyman
Strategies for Implementing Dynamic Load Sharing.
A Local Facility Location Algorithm Supervisor: Assaf Schuster Denis Krivitski Technion – Israel Institute of Technology.
Wide-area cooperative storage with CFS
12006/9/26 Load Balancing in Dynamic Structured P2P Systems Brighten Godfrey, Karthik Lakshminarayanan, Sonesh Surana, Richard Karp, Ion Stoica INFOCOM.
Tapestry: Finding Nearby Objects in Peer-to-Peer Networks Joint with: Ling Huang Anthony Joseph Robert Krauthgamer John Kubiatowicz Satish Rao Sean Rhea.
ICDE A Peer-to-peer Framework for Caching Range Queries Ozgur D. Sahin Abhishek Gupta Divyakant Agrawal Amr El Abbadi Department of Computer Science.
Structured P2P Network Group14: Qiwei Zhang; Shi Yan; Dawei Ouyang; Boyu Sun.
Algorithms for Self-Organization and Adaptive Service Placement in Dynamic Distributed Systems Artur Andrzejak, Sven Graupner,Vadim Kotov, Holger Trinks.
Dynamo: Amazon’s Highly Available Key-value Store Presented By: Devarsh Patel 1CS5204 – Operating Systems.
Distributed Data Stores – Facebook Presented by Ben Gooding University of Arkansas – April 21, 2015.
Towards Efficient Load Balancing in Structured P2P Systems Yingwu Zhu, Yiming Hu University of Cincinnati.
PIC: Practical Internet Coordinates for Distance Estimation Manuel Costa joint work with Miguel Castro, Ant Rowstron, Peter Key Microsoft Research Cambridge.
Thesis Proposal Data Consistency in DHTs. Background Peer-to-peer systems have become increasingly popular Lots of P2P applications around us –File sharing,
Load Balancing in Structured P2P System Ananth Rao, Karthik Lakshminarayanan, Sonesh Surana, Richard Karp, Ion Stoica IPTPS ’03 Kyungmin Cho 2003/05/20.
Load Balancing of In-Network Data-Centric Storage Schemes in Sensor Networks Mohamed Aly In collaboration with Kirk Pruhs and Panos K. Chrysanthis Advanced.
Min Xu1, Yunfeng Zhu2, Patrick P. C. Lee1, Yinlong Xu2
Network Aware Resource Allocation in Distributed Clouds.
Overcast: Reliable Multicasting with an Overlay Network CS294 Paul Burstein 9/15/2003.
© Copyright 2012 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. LogKV: Exploiting Key-Value.
Scalable Web Server on Heterogeneous Cluster CHEN Ge.
ENERGY-EFFICIENT FORWARDING STRATEGIES FOR GEOGRAPHIC ROUTING in LOSSY WIRELESS SENSOR NETWORKS Presented by Prasad D. Karnik.
Fast Crash Recovery in RAMCloud. Motivation The role of DRAM has been increasing – Facebook used 150TB of DRAM For 200TB of disk storage However, there.
1 Secure Peer-to-Peer File Sharing Frans Kaashoek, David Karger, Robert Morris, Ion Stoica, Hari Balakrishnan MIT Laboratory.
Scalable and Topology-Aware Load Balancers in Charm++ Amit Sharma Parallel Programming Lab, UIUC.
Topologically-Aware Overlay Construction and Sever Selection Sylvia Ratnasamy, Mark Handley, Richard Karp, Scott Shenker.
LOOKING UP DATA IN P2P SYSTEMS Hari Balakrishnan M. Frans Kaashoek David Karger Robert Morris Ion Stoica MIT LCS.
PeerNet: Pushing Peer-to-Peer Down the Stack Jakob Eriksson, Michalis Faloutsos, Srikanth Krishnamurthy University of California, Riverside.
Routing in Delay Tolerant Network Qing Ye EDIFY Group of Lehigh University.
CS694 - DHT1 Distributed Hash Table Systems Hui Zhang University of Southern California.
Department of Computer Science, Johns Hopkins University EN Instructor: Randal Burns 24 September 2013 NoSQL Data Models and Systems.
School of Computing Clemson University Fall, 2012
Introduction to SDNS-Mon
Data Driven Resource Allocation for Distributed Learning
Introduction to Load Balancing:
(slides by Nick Feamster)
SCOPE: Scalable Consistency in Structured P2P Systems
Providing Secure Storage on the Internet
DHT Routing Geometries and Chord
MON TUE WED THU
2008 Calendar.
Sun Mon Tue Wed Thu Fri Sat
A Scalable Content Addressable Network
Sun Mon Tue Wed Thu Fri Sat
1/○~1/○ weekly schedule MON TUE WED THU FRI SAT SUN MEMO
2016 | 10 OCT SUN MON TUE WED THU FRI SAT
Sun Mon Tue Wed Thu Fri Sat
2008 Calendar.
Presentation transcript:

Distributed Load Balancing for Key-Value Storage Systems Imranul Hoque Michael Spreitzer Malgorzata Steinder

2

Key-Value Storage Systems Usage: – Session state, tags, comments, etc. Requirements: – Scalability – Fast response time – High availability & fault tolerance – Relaxed consistency guarantee Example: Cassandra, Dynamo, PNUTS, etc. 3

Load Balancing in K-V Storage Hash partitioned vs. range partitioned – Range partitioned data ensures efficient range scan/search – Hash partitioned data helps even distribution Server 1 Server 2 Server 3 Server 4 SAT TUE SUN MON WED THU FRI MON TUE WED THU FRI SAT SUN Tablets Table 4

Issues with Load Balancing Uneven space distribution due to range partitioning – Solution: partition the tablets and move them around Few number of very popular records Server 1 Server 2 Server 3 Server 4 SAT TUE SUN MON WED THU FRI 5

Contribution Algorithms for solving the load balancing problem – Load = space, bandwidth – Evenly distribute the spare capacity – Distributed algorithm, not a centralized one – Reduce the number of moves Previous solutions: – One dimensional/key-space redistribution/bulk loading 6

Outline Motivation System modeling and assumptions Algorithms – One-to-one – One-to-n – Move suppression Design decisions Experimental results Emulation of proposed distributed algorithms Future works 7

System Modeling and Assumptions Table Tablet Server A Server B Server C B 1, S 1 B 2, S 2 B 3, S 3 B A, S A B B, S B B C, S C 8 1.<= 0.01 in both dimensions 2. # of tablets >> # of nodes 1.<= 0.01 in both dimensions 2. # of tablets >> # of nodes B 1, S 1 B 4, S 4 B 5, S 5

System State B B SS Target Zone: helps achieve convergence Target Point Goal: Move tablets around so that every server is within the target zone 9

Load Balancing Algorithms Phase 1: – Global averaging scheme – Variance of the approximation of the average decreases exponentially fast Phase 2: – One-to-one gossip – One-to-n gossip – Move suppression Phase 1 Phase 2 Phase 1 Phase 2 t 10

One-to-One Gossip Point selection strategy – Midpoint strategy – Greedy strategy Tablet transfer strategy – Move to the selected point with minimum cost (space transferred) 11

Tablet Transfer Strategy Server 2 Server 1 Target for Server 1 B B SS 12

Tablet Transfer Strategy (2) Server 1 Left Right Start with an empty bag Goal: take vectors from the servers so that they add up to the target vector If slope(bag + left + right) < slope(target): – Add right to bag, move right – Otherwise, add left to bag move left 13

Initial Configurations Uniform Two Extreme Mid Quadrant 14

Point Selection Strategy Midpoint Strategy + Guaranteed convergence + No need to run phase 1 – Lots of extra movement Visualization Demo – Uniform Uniform – Two extreme Two extreme – Mid quadrant Mid quadrant SS B B Server 1 Server 2 15

Point Selection Strategy (2) Greedy Strategy – Take the point closer to the target – Move it to the target, if improves the position of the other point does not worsen by more than δ Reduces movement Server 1 Server 2 Takes long time to converge in some casessome Takes long time to converge in some casessome 16

DHT-based Location Directory 17

DHT + Midpoint Greedy + fallback to DHT: – Convergence problem exists for some configurations – Visualization Demo Visualization Demo Solution: – Greedy + fallback to DHT with Midpoint – Demo: uniform, two extreme, mid quadrantuniformtwo extrememid quadrant Alternate approach: – Greedy + fallback to Midpoint – Trade-off: movement cost vs. DHT overhead 18

Experimental Evaluation Uniform configuration – Greedy + DHT (Midpoint) – Midpoint – Greedy + Midpoint (No DHT) Effect of varying target zone Effect of failed gossip count Metrics – Amount of space moved – # of gossip rounds – Multiple tablet move 19

Uniform Configuration: Results 20

Effect of Varying Target Zone Larger target zone = fast convergence, less accuracy 21 Target zone width should depend on the target point value

Effect of Failed Gossip Count (Greedy) Large failed gossip count = More time in greedy mode, more unproductive gossip at the end 22

One-to-N Gossip Contact a few random nodes – Locked/unlocked mode Pick the most profitable one – Distance from the target is minimized Advantage – Better choices Initial results – Locked mode: may lead to deadlock – Unlocked mode: most of the cases other nodes start transfer 23

Move Suppression Two global stages Stage 1: – One-to-One gossip, but moves are hypothetical Stage 2: – Change to chosen placement Advantage – Tablet not moved multiple times Challenges – When to switch to Stage 2 from Stage 1 24

Future Works Handling initial placement Frequency of running the placement algorithm Considering the network hierarchy Handling failures Extending to heterogeneous resources Questions? 25