Big data, little time. Scale-out data serving. Scale-out data serving. Highly skewed key popularity

Size: px
Start display at page:

Download "Big data, little time. Scale-out data serving. Scale-out data serving. Highly skewed key popularity"

Transcription

1 /7/6 Big data, little time Goal is to keep (hot) data in memory Requires scale-out approach Each server responsible for one chunk Fast access to local data The Case for RackOut Scalable Data Serving Using Rack-Scale Systems # # Computation/serving performed in parallel Stanko Novakovic, Alexandros Daglis, Edouard Bugnion, Babak Falsafi, Boris Grot # #N Scale-out model offers plenty of DRAM & fast local access Scale-out data serving Scale-out data serving Central to many web applications Central to many web applications e.g., social networks, e-commerce e.g., social networks, e-commerce Data sharded using consistent hashing Data sharded using consistent hashing # # Client pinpoints based on hash # hosts collection of micros = set of key-value pairs #N # hosts collection of micros = set of key-value pairs # # #N Fast data lookup based on client-side consistent hashing 3 Highly skewed key popularity Skewed access distribution Why is skew problematic? Shard skew: skew across servers - Shard_skew = MAX/AVG - Zipfian typically Hottest server saturates while most servers barely utilized Service Level Objective (SLO) violations can occur below that level keys 99th percentile lat. MAX hash(key) λi μ Hundreds AVG Arrival rate (load) Billions load popularity Saturation level servers Skewed popularity translates to load imbalance SLO s SLO limits utilization of DC resources 6

2 /7/6 How do we deal with skew today? Dynamic migration/replication Dynamic migration & replication techniques Dynamic replication is a trade-off that requires: - Monitor load and detect load bursts - te/migrate hot micro(s) - Balance load across replicas Rack #N LOAD [Huang ] - s of replicas in social networking workloads - Load monitoring, copying/moving data & metadata updates Higher skew translates to more replicas Dynamic replication & migration require: # # # - Additional memory & consistency model replica #N Hot micro Mitigate load imbalance using migration or multiple replicas Higher skew à higher replication overhead! 7 Insight: fewer nodes results in smaller skew Contributions/Outline Scaling in contrary to scaling out reduces shard skew Analysis of load imbalance in data serving - Better load distribution - Fewer replicas needed Popularity skew translates to replication overhead RackOut: A technique for mitigating load imbalance Load ~% of total load (shard_skew = 3) Experimental methodology Combination of queuing model and real implementation s ~% of total load (shard_skew = 3) Load Reduces replication overhead s 3 [ASPLOS ] Evaluation using RDMA and Scale-Out NUMA Fewer data shards results in less imbalance, less overhead 9 RackOut rather than scale-out Scale nodes to host more keys and absorb higher load - e.g., 6x fewer nodes à 6x more memory per node - scale to the size of a rack? Scale-out Super RackOut 99th percentile lat. What if we could make nodes larger? SLO RackOut improves throughput w/ no replication cost

3 N /7/6 Towards rack-scale building blocks Scaling up shared memory is expensive Cost & complexity of HW cache coherence, fault containment Remote Direct Memory Access (RDMA) Enables low-latency access to remote memory Hardware transport, destination CPU not involved e.g. Infiniband over IP and lossless Ethernet (RoCE) Is RDMA is the new scale-up? Extreme case: Full-scale RDMA DC-scale RoCE introduces emergent safety & perf. issues: - PFC-induced congestion - PFC deadlock, - RDMA transport livelock, - pause frame storm, - etc [Zhu, Guo 6] TCP/IP RDMA 3 A hybrid approach: RackOut using RDMA TCP/IP... A hybrid approach: RackOut using RDMA Sweet spot between scale-out and full-scale RDMA - Enable sharing within rack ( super ) RackOut follows scale-out model - Clients connect via network - Consistent hashing - Migration & replication still possible - Only across racks RDMA RDMA RDMA Grouping Factor (GF) defines the size of a rack Reduce imbalance by using RDMA rack as building block Concurrent Exclusive (CREW) Random Select node Request R/W Previously introduced for multicores Shared X Owner of X [Lim ] Shared read-only access to data, exclusive writes 6 Client-rack architecture of RackOut CREW enables load balancing of read operations - tion only across racks à Rack Ū [..GF] Rand S # # RDMA Rack #N Enable load balancing using RDMA via CREW #N 7 3

4 /7/6 Methodology Contributions/Outline Queuing model for modeling DC-scale RackOut Analysis of load imbalance in data serving Input: node count, GF, read-write ratio, distribution, etc. Popularity skew translates to replication overhead A RackOut KVS implementation (RO-KVS) RackOut: A technique for mitigating load imbalance Reduces replication overhead. Instrument model using platform s parameters. Validate model w/ actual measurements 3. Use model to evaluate arbitrary configurations Experimental methodology Combination of queuing model and real implementation [ASPLOS ] Evaluation using RDMA and Scale-Out NUMA 9 RackOut KVS (RO-KVS) Queuing model for RackOut [Dragojevic, ] Uses FaRM framework as foundation Discrete event-based simulation Poisson process, three service times, Zipfian Client Ū [..GF] Rand S S GF Poisson ( λ ) Optimistic Concurrency Control (OCC) [ASPLOS ] Runs on Mellanox RoCE and Scale-Out NUMA Rack (N/GF) S S GF R/W S Key S Rack. Ū [9%/%] Hash space divided among servers (in micros) s can read from all micros within their rack LR RR LW à Rack α) Zipf ( Both clients & servers maintain DHT of cluster size s -benchmarks for measuring service times YCSB workloads with skewed distributions Local (LR); Remote (RR); Local (LW) Fast & accurate modeling of arbitrary RackOut confs Full-scale RackOut DC simulation Model validation (hottest rack) YCSB-B workload (% writes) on hottest group of 6 servers Modeling -node datacenter (YCSB-B) Dashed lines show platform results GF GF GF GF GF6 99th-pct latency (ms) GF GF GF GF6 99th-pct latency (ms) GF 3 6 (% of max. capacity) 7 Rack throughput (% of max. capacity) Model provides accurate RackOut evaluation (<6% error) Model (GF) RO-KVS (GF) Model (GF) RO-KVS (GF) Model (GF) RO-KVS (GF) Model (GF) RO-KVS (GF) Model (GF6) RO-KVS (GF6) RackOut improves TPS w/o violating SLO at DC scale

5 /7/6 RackOut is synergistic w/ replication Greedy dynamic migration and replication algorithm Accounts for consistent updates (% of max. capacity) 3 3 GF GF GF GF GF6 6 Number of replications tion consumes less resources w/ RackOut Sensitivity to remote latency Lower RR/LR ratio à higher impact of RackOut Speedup (over GF) GF GF GF GF GF6 sonuma RoCE RR/LR % higher speedup with sonuma because of lower RR 6 7 Conclusion RackOut: an approach to scaling in using RDMA RackOut reduces skew and replication overheads Requires fewer replicas to absorb high skew 7 RackOut platforms. Intel Xeon E with Mellanox ConnectX-3 (RoCE) Mellanox RDMA RackOut platform Intel Xeon E with Mellanox RDMA (RoCE). VMM-based sonuma emulator outstanding (latency) Obtain service times for RackOut Local, Remote, Local RO-KVS super Soft RMC Soft RMC FaRM FaRM SRIOV Intel x YCSB (coordinator) Network switch (GbE) Aggregate TPS YCSB (load generators) x LR = us; RR =.us; LW = 6.9us sonuma rack (Xen VMM) N outstanding (TPS) 3 9

Be Fast, Cheap and in Control with SwitchKV. Xiaozhou Li

Be Fast, Cheap and in Control with SwitchKV. Xiaozhou Li Be Fast, Cheap and in Control with SwitchKV Xiaozhou Li Goal: fast and cost-efficient key-value store Store, retrieve, manage key-value objects Get(key)/Put(key,value)/Delete(key) Target: cluster-level

More information

Be Fast, Cheap and in Control with SwitchKV Xiaozhou Li

Be Fast, Cheap and in Control with SwitchKV Xiaozhou Li Be Fast, Cheap and in Control with SwitchKV Xiaozhou Li Raghav Sethi Michael Kaminsky David G. Andersen Michael J. Freedman Goal: fast and cost-effective key-value store Target: cluster-level storage for

More information

Revisiting Network Support for RDMA

Revisiting Network Support for RDMA Revisiting Network Support for RDMA Radhika Mittal 1, Alex Shpiner 3, Aurojit Panda 1, Eitan Zahavi 3, Arvind Krishnamurthy 2, Sylvia Ratnasamy 1, Scott Shenker 1 (1: UC Berkeley, 2: Univ. of Washington,

More information

Red Hat Gluster Storage performance. Manoj Pillai and Ben England Performance Engineering June 25, 2015

Red Hat Gluster Storage performance. Manoj Pillai and Ben England Performance Engineering June 25, 2015 Red Hat Gluster Storage performance Manoj Pillai and Ben England Performance Engineering June 25, 2015 RDMA Erasure Coding NFS-Ganesha New or improved features (in last year) Snapshots SSD support Erasure

More information

Low-Overhead Flash Disaggregation via NVMe-over-Fabrics Vijay Balakrishnan Memory Solutions Lab. Samsung Semiconductor, Inc.

Low-Overhead Flash Disaggregation via NVMe-over-Fabrics Vijay Balakrishnan Memory Solutions Lab. Samsung Semiconductor, Inc. Low-Overhead Flash Disaggregation via NVMe-over-Fabrics Vijay Balakrishnan Memory Solutions Lab. Samsung Semiconductor, Inc. 1 DISCLAIMER This presentation and/or accompanying oral statements by Samsung

More information

Low-Overhead Flash Disaggregation via NVMe-over-Fabrics

Low-Overhead Flash Disaggregation via NVMe-over-Fabrics Low-Overhead Flash Disaggregation via NVMe-over-Fabrics Vijay Balakrishnan Memory Solutions Lab. Samsung Semiconductor, Inc. August 2017 1 DISCLAIMER This presentation and/or accompanying oral statements

More information

Scale-Out ccnuma: Exploiting Skew with Strongly Consistent Caching

Scale-Out ccnuma: Exploiting Skew with Strongly Consistent Caching Scale-Out ccnuma: Exploiting Skew with Strongly Consistent Caching Vasilis Gavrielatos The University of Edinburgh Vasilis.Gavrielatos@ed.ac.uk Nicolai Oswald The University of Edinburgh Nicolai.Oswald@ed.ac.uk

More information

High-Performance Key-Value Store on OpenSHMEM

High-Performance Key-Value Store on OpenSHMEM High-Performance Key-Value Store on OpenSHMEM Huansong Fu*, Manjunath Gorentla Venkata, Ahana Roy Choudhury*, Neena Imam, Weikuan Yu* *Florida State University Oak Ridge National Laboratory Outline Background

More information

FaSST: Fast, Scalable, and Simple Distributed Transactions with Two-Sided (RDMA) Datagram RPCs

FaSST: Fast, Scalable, and Simple Distributed Transactions with Two-Sided (RDMA) Datagram RPCs FaSST: Fast, Scalable, and Simple Distributed Transactions with Two-Sided (RDMA) Datagram RPCs Anuj Kalia (CMU), Michael Kaminsky (Intel Labs), David Andersen (CMU) RDMA RDMA is a network feature that

More information

A Distributed Hash Table for Shared Memory

A Distributed Hash Table for Shared Memory A Distributed Hash Table for Shared Memory Wytse Oortwijn Formal Methods and Tools, University of Twente August 31, 2015 Wytse Oortwijn (Formal Methods and Tools, AUniversity Distributed of Twente) Hash

More information

The Google File System

The Google File System The Google File System Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung Google SOSP 03, October 19 22, 2003, New York, USA Hyeon-Gyu Lee, and Yeong-Jae Woo Memory & Storage Architecture Lab. School

More information

Designing Next-Generation Data- Centers with Advanced Communication Protocols and Systems Services. Presented by: Jitong Chen

Designing Next-Generation Data- Centers with Advanced Communication Protocols and Systems Services. Presented by: Jitong Chen Designing Next-Generation Data- Centers with Advanced Communication Protocols and Systems Services Presented by: Jitong Chen Outline Architecture of Web-based Data Center Three-Stage framework to benefit

More information

Designing Next Generation FS for NVMe and NVMe-oF

Designing Next Generation FS for NVMe and NVMe-oF Designing Next Generation FS for NVMe and NVMe-oF Liran Zvibel CTO, Co-founder Weka.IO @liranzvibel Santa Clara, CA 1 Designing Next Generation FS for NVMe and NVMe-oF Liran Zvibel CTO, Co-founder Weka.IO

More information

Advanced RDMA-based Admission Control for Modern Data-Centers

Advanced RDMA-based Admission Control for Modern Data-Centers Advanced RDMA-based Admission Control for Modern Data-Centers Ping Lai Sundeep Narravula Karthikeyan Vaidyanathan Dhabaleswar. K. Panda Computer Science & Engineering Department Ohio State University Outline

More information

CA485 Ray Walshe Google File System

CA485 Ray Walshe Google File System Google File System Overview Google File System is scalable, distributed file system on inexpensive commodity hardware that provides: Fault Tolerance File system runs on hundreds or thousands of storage

More information

FaRM: Fast Remote Memory

FaRM: Fast Remote Memory FaRM: Fast Remote Memory Problem Context DRAM prices have decreased significantly Cost effective to build commodity servers w/hundreds of GBs E.g. - cluster with 100 machines can hold tens of TBs of main

More information

Huge market -- essentially all high performance databases work this way

Huge market -- essentially all high performance databases work this way 11/5/2017 Lecture 16 -- Parallel & Distributed Databases Parallel/distributed databases: goal provide exactly the same API (SQL) and abstractions (relational tables), but partition data across a bunch

More information

BCStore: Bandwidth-Efficient In-memory KV-Store with Batch Coding. Shenglong Li, Quanlu Zhang, Zhi Yang and Yafei Dai Peking University

BCStore: Bandwidth-Efficient In-memory KV-Store with Batch Coding. Shenglong Li, Quanlu Zhang, Zhi Yang and Yafei Dai Peking University BCStore: Bandwidth-Efficient In-memory KV-Store with Batch Coding Shenglong Li, Quanlu Zhang, Zhi Yang and Yafei Dai Peking University Outline Introduction and Motivation Our Design System and Implementation

More information

Lenovo - Excelero NVMesh Reference Architecture

Lenovo - Excelero NVMesh Reference Architecture Lenovo - Excelero NVMesh Reference Architecture How adding a dash of software to your server platform turns DAS into a high performance shared storage solution. Introduction Following the example of Tech

More information

Architecture of a Real-Time Operational DBMS

Architecture of a Real-Time Operational DBMS Architecture of a Real-Time Operational DBMS Srini V. Srinivasan Founder, Chief Development Officer Aerospike CMG India Keynote Thane December 3, 2016 [ CMGI Keynote, Thane, India. 2016 Aerospike Inc.

More information

RoCE vs. iwarp Competitive Analysis

RoCE vs. iwarp Competitive Analysis WHITE PAPER February 217 RoCE vs. iwarp Competitive Analysis Executive Summary...1 RoCE s Advantages over iwarp...1 Performance and Benchmark Examples...3 Best Performance for Virtualization...5 Summary...6

More information

Google File System. Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung Google fall DIP Heerak lim, Donghun Koo

Google File System. Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung Google fall DIP Heerak lim, Donghun Koo Google File System Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung Google 2017 fall DIP Heerak lim, Donghun Koo 1 Agenda Introduction Design overview Systems interactions Master operation Fault tolerance

More information

RDMA over Commodity Ethernet at Scale

RDMA over Commodity Ethernet at Scale RDMA over Commodity Ethernet at Scale Chuanxiong Guo, Haitao Wu, Zhong Deng, Gaurav Soni, Jianxi Ye, Jitendra Padhye, Marina Lipshteyn ACM SIGCOMM 2016 August 24 2016 Outline RDMA/RoCEv2 background DSCP-based

More information

Load Sharing in Peer-to-Peer Networks using Dynamic Replication

Load Sharing in Peer-to-Peer Networks using Dynamic Replication Load Sharing in Peer-to-Peer Networks using Dynamic Replication S Rajasekhar, B Rong, K Y Lai, I Khalil and Z Tari School of Computer Science and Information Technology RMIT University, Melbourne 3, Australia

More information

Infiniswap. Efficient Memory Disaggregation. Mosharaf Chowdhury. with Juncheng Gu, Youngmoon Lee, Yiwen Zhang, and Kang G. Shin

Infiniswap. Efficient Memory Disaggregation. Mosharaf Chowdhury. with Juncheng Gu, Youngmoon Lee, Yiwen Zhang, and Kang G. Shin Infiniswap Efficient Memory Disaggregation Mosharaf Chowdhury with Juncheng Gu, Youngmoon Lee, Yiwen Zhang, and Kang G. Shin Rack-Scale Computing Datacenter-Scale Computing Geo-Distributed Computing Coflow

More information

Highly Scalable, Non-RDMA NVMe Fabric. Bob Hansen,, VP System Architecture

Highly Scalable, Non-RDMA NVMe Fabric. Bob Hansen,, VP System Architecture A Cost Effective,, High g Performance,, Highly Scalable, Non-RDMA NVMe Fabric Bob Hansen,, VP System Architecture bob@apeirondata.com Storage Developers Conference, September 2015 Agenda 3 rd Platform

More information

InfiniBand Networked Flash Storage

InfiniBand Networked Flash Storage InfiniBand Networked Flash Storage Superior Performance, Efficiency and Scalability Motti Beck Director Enterprise Market Development, Mellanox Technologies Flash Memory Summit 2016 Santa Clara, CA 1 17PB

More information

Deconstructing RDMA-enabled Distributed Transaction Processing: Hybrid is Better!

Deconstructing RDMA-enabled Distributed Transaction Processing: Hybrid is Better! Deconstructing RDMA-enabled Distributed Transaction Processing: Hybrid is Better! Xingda Wei, Zhiyuan Dong, Rong Chen, Haibo Chen Institute of Parallel and Distributed Systems (IPADS) Shanghai Jiao Tong

More information

Supporting Strong Cache Coherency for Active Caches in Multi-Tier Data-Centers over InfiniBand

Supporting Strong Cache Coherency for Active Caches in Multi-Tier Data-Centers over InfiniBand Supporting Strong Cache Coherency for Active Caches in Multi-Tier Data-Centers over InfiniBand S. Narravula, P. Balaji, K. Vaidyanathan, S. Krishnamoorthy, J. Wu and D. K. Panda The Ohio State University

More information

E-Store: Fine-Grained Elastic Partitioning for Distributed Transaction Processing Systems

E-Store: Fine-Grained Elastic Partitioning for Distributed Transaction Processing Systems E-Store: Fine-Grained Elastic Partitioning for Distributed Transaction Processing Systems Rebecca Taft, Essam Mansour, Marco Serafini, Jennie Duggan, Aaron J. Elmore, Ashraf Aboulnaga, Andrew Pavlo, Michael

More information

Building Consistent Transactions with Inconsistent Replication

Building Consistent Transactions with Inconsistent Replication Building Consistent Transactions with Inconsistent Replication Irene Zhang, Naveen Kr. Sharma, Adriana Szekeres, Arvind Krishnamurthy, Dan R. K. Ports University of Washington Distributed storage systems

More information

Near Memory Key/Value Lookup Acceleration MemSys 2017

Near Memory Key/Value Lookup Acceleration MemSys 2017 Near Key/Value Lookup Acceleration MemSys 2017 October 3, 2017 Scott Lloyd, Maya Gokhale Center for Applied Scientific Computing This work was performed under the auspices of the U.S. Department of Energy

More information

N V M e o v e r F a b r i c s -

N V M e o v e r F a b r i c s - N V M e o v e r F a b r i c s - H i g h p e r f o r m a n c e S S D s n e t w o r k e d f o r c o m p o s a b l e i n f r a s t r u c t u r e Rob Davis, VP Storage Technology, Mellanox OCP Evolution Server

More information

Implementing SQL Server 2016 with Microsoft Storage Spaces Direct on Dell EMC PowerEdge R730xd

Implementing SQL Server 2016 with Microsoft Storage Spaces Direct on Dell EMC PowerEdge R730xd Implementing SQL Server 2016 with Microsoft Storage Spaces Direct on Dell EMC PowerEdge R730xd Performance Study Dell EMC Engineering October 2017 A Dell EMC Performance Study Revisions Date October 2017

More information

Improving Altibase Performance with Solarflare 10GbE Server Adapters and OpenOnload

Improving Altibase Performance with Solarflare 10GbE Server Adapters and OpenOnload Improving Altibase Performance with Solarflare 10GbE Server Adapters and OpenOnload Summary As today s corporations process more and more data, the business ramifications of faster and more resilient database

More information

No Tradeoff Low Latency + High Efficiency

No Tradeoff Low Latency + High Efficiency No Tradeoff Low Latency + High Efficiency Christos Kozyrakis http://mast.stanford.edu Latency-critical Applications A growing class of online workloads Search, social networking, software-as-service (SaaS),

More information

High Performance Transactions in Deuteronomy

High Performance Transactions in Deuteronomy High Performance Transactions in Deuteronomy Justin Levandoski, David Lomet, Sudipta Sengupta, Ryan Stutsman, and Rui Wang Microsoft Research Overview Deuteronomy: componentized DB stack Separates transaction,

More information

Meltdown and Spectre Interconnect Performance Evaluation Jan Mellanox Technologies

Meltdown and Spectre Interconnect Performance Evaluation Jan Mellanox Technologies Meltdown and Spectre Interconnect Evaluation Jan 2018 1 Meltdown and Spectre - Background Most modern processors perform speculative execution This speculation can be measured, disclosing information about

More information

Distributed Shared Memory Consistency Object-based Model

Distributed Shared Memory Consistency Object-based Model Journal of Computer Science 3 (1): 57-61, 27 ISSN1549-3636 27 Science Publications Corresponding Author: Distributed Shared Memory Consistency Object-based Model Abdelfatah Aref Yahya and Rana Mohamad

More information

Designing Distributed Systems using Approximate Synchrony in Data Center Networks

Designing Distributed Systems using Approximate Synchrony in Data Center Networks Designing Distributed Systems using Approximate Synchrony in Data Center Networks Dan R. K. Ports Jialin Li Naveen Kr. Sharma Vincent Liu Arvind Krishnamurthy University of Washington CSE Today s most

More information

S. Narravula, P. Balaji, K. Vaidyanathan, H.-W. Jin and D. K. Panda. The Ohio State University

S. Narravula, P. Balaji, K. Vaidyanathan, H.-W. Jin and D. K. Panda. The Ohio State University Architecture for Caching Responses with Multiple Dynamic Dependencies in Multi-Tier Data- Centers over InfiniBand S. Narravula, P. Balaji, K. Vaidyanathan, H.-W. Jin and D. K. Panda The Ohio State University

More information

Reducing Network Contention with Mixed Workloads on Modern Multicore Clusters

Reducing Network Contention with Mixed Workloads on Modern Multicore Clusters Reducing Network Contention with Mixed Workloads on Modern Multicore Clusters Matthew Koop 1 Miao Luo D. K. Panda matthew.koop@nasa.gov {luom, panda}@cse.ohio-state.edu 1 NASA Center for Computational

More information

Ambry: LinkedIn s Scalable Geo- Distributed Object Store

Ambry: LinkedIn s Scalable Geo- Distributed Object Store Ambry: LinkedIn s Scalable Geo- Distributed Object Store Shadi A. Noghabi *, Sriram Subramanian +, Priyesh Narayanan +, Sivabalan Narayanan +, Gopalakrishna Holla +, Mammad Zadeh +, Tianwei Li +, Indranil

More information

Benefits of 25, 40, and 50GbE Networks for Ceph and Hyper- Converged Infrastructure John F. Kim Mellanox Technologies

Benefits of 25, 40, and 50GbE Networks for Ceph and Hyper- Converged Infrastructure John F. Kim Mellanox Technologies Benefits of 25, 40, and 50GbE Networks for Ceph and Hyper- Converged Infrastructure John F. Kim Mellanox Technologies Storage Transitions Change Network Needs Software Defined Storage Flash Storage Storage

More information

CSE 124: Networked Services Lecture-17

CSE 124: Networked Services Lecture-17 Fall 2010 CSE 124: Networked Services Lecture-17 Instructor: B. S. Manoj, Ph.D http://cseweb.ucsd.edu/classes/fa10/cse124 11/30/2010 CSE 124 Networked Services Fall 2010 1 Updates PlanetLab experiments

More information

Accelerating Big Data: Using SanDisk SSDs for Apache HBase Workloads

Accelerating Big Data: Using SanDisk SSDs for Apache HBase Workloads WHITE PAPER Accelerating Big Data: Using SanDisk SSDs for Apache HBase Workloads December 2014 Western Digital Technologies, Inc. 951 SanDisk Drive, Milpitas, CA 95035 www.sandisk.com Table of Contents

More information

FaRM: Fast Remote Memory

FaRM: Fast Remote Memory FaRM: Fast Remote Memory Aleksandar Dragojević, Dushyanth Narayanan, Orion Hodson, Miguel Castro Microsoft Research Abstract We describe the design and implementation of FaRM, a new main memory distributed

More information

Tales of the Tail Hardware, OS, and Application-level Sources of Tail Latency

Tales of the Tail Hardware, OS, and Application-level Sources of Tail Latency Tales of the Tail Hardware, OS, and Application-level Sources of Tail Latency Jialin Li, Naveen Kr. Sharma, Dan R. K. Ports and Steven D. Gribble February 2, 2015 1 Introduction What is Tail Latency? What

More information

Chelsio Communications. Meeting Today s Datacenter Challenges. Produced by Tabor Custom Publishing in conjunction with: CUSTOM PUBLISHING

Chelsio Communications. Meeting Today s Datacenter Challenges. Produced by Tabor Custom Publishing in conjunction with: CUSTOM PUBLISHING Meeting Today s Datacenter Challenges Produced by Tabor Custom Publishing in conjunction with: 1 Introduction In this era of Big Data, today s HPC systems are faced with unprecedented growth in the complexity

More information

vsan 6.6 Performance Improvements First Published On: Last Updated On:

vsan 6.6 Performance Improvements First Published On: Last Updated On: vsan 6.6 Performance Improvements First Published On: 07-24-2017 Last Updated On: 07-28-2017 1 Table of Contents 1. Overview 1.1.Executive Summary 1.2.Introduction 2. vsan Testing Configuration and Conditions

More information

Data Processing on Emerging Hardware

Data Processing on Emerging Hardware Data Processing on Emerging Hardware Gustavo Alonso Systems Group Department of Computer Science ETH Zurich, Switzerland 3 rd International Summer School on Big Data, Munich, Germany, 2017 www.systems.ethz.ch

More information

Audience This paper is targeted for IT managers and architects. It showcases how to utilize your network efficiently and gain higher performance using

Audience This paper is targeted for IT managers and architects. It showcases how to utilize your network efficiently and gain higher performance using White paper Benefits of Remote Direct Memory Access Over Routed Fabrics Introduction An enormous impact on data center design and operations is happening because of the rapid evolution of enterprise IT.

More information

SR-IOV Support for Virtualization on InfiniBand Clusters: Early Experience

SR-IOV Support for Virtualization on InfiniBand Clusters: Early Experience SR-IOV Support for Virtualization on InfiniBand Clusters: Early Experience Jithin Jose, Mingzhe Li, Xiaoyi Lu, Krishna Kandalla, Mark Arnold and Dhabaleswar K. (DK) Panda Network-Based Computing Laboratory

More information

Elastic Scaling of Stateful Network Functions

Elastic Scaling of Stateful Network Functions NSDI 2018 Elastic Scaling of Stateful Network Functions Shinae Woo *+, Justine Sherry *, Sangjin Han *, Sue Moon +, Sylvia Ratnasamy *, Scott Shenker * + KAIST, * UC Berkeley Elastic Scaling of NFs NFV

More information

Jinho Hwang (IBM Research) Wei Zhang, Timothy Wood, H. Howie Huang (George Washington Univ.) K.K. Ramakrishnan (Rutgers University)

Jinho Hwang (IBM Research) Wei Zhang, Timothy Wood, H. Howie Huang (George Washington Univ.) K.K. Ramakrishnan (Rutgers University) Jinho Hwang (IBM Research) Wei Zhang, Timothy Wood, H. Howie Huang (George Washington Univ.) K.K. Ramakrishnan (Rutgers University) Background: Memory Caching Two orders of magnitude more reads than writes

More information

Alexandros Daglis Computer Science PhD Candidate École Polytechnique Fédérale de Lausanne (EPFL)

Alexandros Daglis Computer Science PhD Candidate École Polytechnique Fédérale de Lausanne (EPFL) Alexandros Daglis Computer Science PhD Candidate École Polytechnique Fédérale de Lausanne (EPFL) Computer & Communication Sciences +41 21 69 31385 École Polytechnique Fédérale de Lausanne http://parsa.epfl.ch/

More information

On Smart Query Routing: For Distributed Graph Querying with Decoupled Storage

On Smart Query Routing: For Distributed Graph Querying with Decoupled Storage On Smart Query Routing: For Distributed Graph Querying with Decoupled Storage Arijit Khan Nanyang Technological University (NTU), Singapore Gustavo Segovia ETH Zurich, Switzerland Donald Kossmann Microsoft

More information

The Google File System

The Google File System The Google File System Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung Google* 정학수, 최주영 1 Outline Introduction Design Overview System Interactions Master Operation Fault Tolerance and Diagnosis Conclusions

More information

ASPERA HIGH-SPEED TRANSFER. Moving the world s data at maximum speed

ASPERA HIGH-SPEED TRANSFER. Moving the world s data at maximum speed ASPERA HIGH-SPEED TRANSFER Moving the world s data at maximum speed ASPERA HIGH-SPEED FILE TRANSFER 80 GBIT/S OVER IP USING DPDK Performance, Code, and Architecture Charles Shiflett Developer of next-generation

More information

Spark Over RDMA: Accelerate Big Data SC Asia 2018 Ido Shamay Mellanox Technologies

Spark Over RDMA: Accelerate Big Data SC Asia 2018 Ido Shamay Mellanox Technologies Spark Over RDMA: Accelerate Big Data SC Asia 2018 Ido Shamay 1 Apache Spark - Intro Spark within the Big Data ecosystem Data Sources Data Acquisition / ETL Data Storage Data Analysis / ML Serving 3 Apache

More information

Memcached Design on High Performance RDMA Capable Interconnects

Memcached Design on High Performance RDMA Capable Interconnects Memcached Design on High Performance RDMA Capable Interconnects Jithin Jose, Hari Subramoni, Miao Luo, Minjia Zhang, Jian Huang, Md. Wasi- ur- Rahman, Nusrat S. Islam, Xiangyong Ouyang, Hao Wang, Sayantan

More information

The Google File System

The Google File System The Google File System By Ghemawat, Gobioff and Leung Outline Overview Assumption Design of GFS System Interactions Master Operations Fault Tolerance Measurements Overview GFS: Scalable distributed file

More information

Tailwind: Fast and Atomic RDMA-based Replication. Yacine Taleb, Ryan Stutsman, Gabriel Antoniu, Toni Cortes

Tailwind: Fast and Atomic RDMA-based Replication. Yacine Taleb, Ryan Stutsman, Gabriel Antoniu, Toni Cortes Tailwind: Fast and Atomic RDMA-based Replication Yacine Taleb, Ryan Stutsman, Gabriel Antoniu, Toni Cortes In-Memory Key-Value Stores General purpose in-memory key-value stores are widely used nowadays

More information

NetCache: Balancing Key-Value Stores with Fast In-Network Caching

NetCache: Balancing Key-Value Stores with Fast In-Network Caching NetCache: Balancing Key-Value Stores with Fast In-Network Caching Xin Jin, Xiaozhou Li, Haoyu Zhang, Robert Soulé Jeongkeun Lee, Nate Foster, Changhoon Kim, Ion Stoica NetCache is a rack-scale key-value

More information

NetCache: Balancing Key-Value Stores with Fast In-Network Caching

NetCache: Balancing Key-Value Stores with Fast In-Network Caching NetCache: Balancing Key-Value Stores with Fast In-Network Caching Xin Jin, Xiaozhou Li, Haoyu Zhang, Robert Soulé Jeongkeun Lee, Nate Foster, Changhoon Kim, Ion Stoica NetCache is a rack-scale key-value

More information

SNAP Performance Benchmark and Profiling. April 2014

SNAP Performance Benchmark and Profiling. April 2014 SNAP Performance Benchmark and Profiling April 2014 Note The following research was performed under the HPC Advisory Council activities Participating vendors: HP, Mellanox For more information on the supporting

More information

NPTEL Course Jan K. Gopinath Indian Institute of Science

NPTEL Course Jan K. Gopinath Indian Institute of Science Storage Systems NPTEL Course Jan 2012 (Lecture 39) K. Gopinath Indian Institute of Science Google File System Non-Posix scalable distr file system for large distr dataintensive applications performance,

More information

Oracle TimesTen Scaleout: Revolutionizing In-Memory Transaction Processing

Oracle TimesTen Scaleout: Revolutionizing In-Memory Transaction Processing Oracle Scaleout: Revolutionizing In-Memory Transaction Processing Scaleout is a brand new, shared nothing scale-out in-memory database designed for next generation extreme OLTP workloads. Featuring elastic

More information

Lecture 15: Datacenter TCP"

Lecture 15: Datacenter TCP Lecture 15: Datacenter TCP" CSE 222A: Computer Communication Networks Alex C. Snoeren Thanks: Mohammad Alizadeh Lecture 15 Overview" Datacenter workload discussion DC-TCP Overview 2 Datacenter Review"

More information

Database Replication in Tashkent. CSEP 545 Transaction Processing Sameh Elnikety

Database Replication in Tashkent. CSEP 545 Transaction Processing Sameh Elnikety Database Replication in Tashkent CSEP 545 Transaction Processing Sameh Elnikety Replication for Performance Expensive Limited scalability DB Replication is Challenging Single database system Large, persistent

More information

Forget IOPS: A Proper Way to Characterize & Test Storage Performance Peter Murray SwiftTest

Forget IOPS: A Proper Way to Characterize & Test Storage Performance Peter Murray SwiftTest Forget IOPS: A Proper Way to Characterize & Test Storage Performance Peter Murray peter@swifttest.com SwiftTest Storage Performance Validation Rely on vendor IOPS claims Test in production and pray Validate

More information

Sub-millisecond Stateful Stream Querying over Fast-evolving Linked Data

Sub-millisecond Stateful Stream Querying over Fast-evolving Linked Data Sub-millisecond Stateful Stream Querying over Fast-evolving Linked Data Yunhao Zhang, Rong Chen, Haibo Chen Institute of Parallel and Distributed Systems (IPADS) Shanghai Jiao Tong University Stream Query

More information

TIBCO, HP and Mellanox High Performance Extreme Low Latency Messaging

TIBCO, HP and Mellanox High Performance Extreme Low Latency Messaging TIBCO, HP and Mellanox High Performance Extreme Low Latency Messaging Executive Summary: With the recent release of TIBCO FTL TM, TIBCO is once again changing the game when it comes to providing high performance

More information

Today CSCI Coda. Naming: Volumes. Coda GFS PAST. Instructor: Abhishek Chandra. Main Goals: Volume is a subtree in the naming space

Today CSCI Coda. Naming: Volumes. Coda GFS PAST. Instructor: Abhishek Chandra. Main Goals: Volume is a subtree in the naming space Today CSCI 5105 Coda GFS PAST Instructor: Abhishek Chandra 2 Coda Main Goals: Availability: Work in the presence of disconnection Scalability: Support large number of users Successor of Andrew File System

More information

Apache Spark Graph Performance with Memory1. February Page 1 of 13

Apache Spark Graph Performance with Memory1. February Page 1 of 13 Apache Spark Graph Performance with Memory1 February 2017 Page 1 of 13 Abstract Apache Spark is a powerful open source distributed computing platform focused on high speed, large scale data processing

More information

Understanding Data Locality in VMware vsan First Published On: Last Updated On:

Understanding Data Locality in VMware vsan First Published On: Last Updated On: Understanding Data Locality in VMware vsan First Published On: 07-20-2016 Last Updated On: 09-30-2016 1 Table of Contents 1. Understanding Data Locality in VMware vsan 1.1.Introduction 1.2.vSAN Design

More information

CS533 Concepts of Operating Systems. Jonathan Walpole

CS533 Concepts of Operating Systems. Jonathan Walpole CS533 Concepts of Operating Systems Jonathan Walpole Disco : Running Commodity Operating Systems on Scalable Multiprocessors Outline Goal Problems and solutions Virtual Machine Monitors(VMM) Disco architecture

More information

Accelerating Data Science. Gustavo Alonso Systems Group Department of Computer Science ETH Zurich, Switzerland

Accelerating Data Science. Gustavo Alonso Systems Group Department of Computer Science ETH Zurich, Switzerland Accelerating Data Science Gustavo Alonso Systems Group Department of Computer Science ETH Zurich, Switzerland Data processing today: Appliances (large machines) Data Centers (many machines) Databases are

More information

Systems Infrastructure for Data Science. Web Science Group Uni Freiburg WS 2014/15

Systems Infrastructure for Data Science. Web Science Group Uni Freiburg WS 2014/15 Systems Infrastructure for Data Science Web Science Group Uni Freiburg WS 2014/15 Lecture X: Parallel Databases Topics Motivation and Goals Architectures Data placement Query processing Load balancing

More information

Next-Generation NVMe-Native Parallel Filesystem for Accelerating HPC Workloads

Next-Generation NVMe-Native Parallel Filesystem for Accelerating HPC Workloads Next-Generation NVMe-Native Parallel Filesystem for Accelerating HPC Workloads Liran Zvibel CEO, Co-founder WekaIO @liranzvibel 1 WekaIO Matrix: Full-featured and Flexible Public or Private S3 Compatible

More information

The Google File System

The Google File System The Google File System Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung SOSP 2003 presented by Kun Suo Outline GFS Background, Concepts and Key words Example of GFS Operations Some optimizations in

More information

GFS: The Google File System. Dr. Yingwu Zhu

GFS: The Google File System. Dr. Yingwu Zhu GFS: The Google File System Dr. Yingwu Zhu Motivating Application: Google Crawl the whole web Store it all on one big disk Process users searches on one big CPU More storage, CPU required than one PC can

More information

DataON and Intel Select Hyper-Converged Infrastructure (HCI) Maximizes IOPS Performance for Windows Server Software-Defined Storage

DataON and Intel Select Hyper-Converged Infrastructure (HCI) Maximizes IOPS Performance for Windows Server Software-Defined Storage Solution Brief DataON and Intel Select Hyper-Converged Infrastructure (HCI) Maximizes IOPS Performance for Windows Server Software-Defined Storage DataON Next-Generation All NVMe SSD Flash-Based Hyper-Converged

More information

Oracle Database 12c: JMS Sharded Queues

Oracle Database 12c: JMS Sharded Queues Oracle Database 12c: JMS Sharded Queues For high performance, scalable Advanced Queuing ORACLE WHITE PAPER MARCH 2015 Table of Contents Introduction 2 Architecture 3 PERFORMANCE OF AQ-JMS QUEUES 4 PERFORMANCE

More information

Inexpensive Coordination in Hardware

Inexpensive Coordination in Hardware Consensus in a Box Inexpensive Coordination in Hardware Zsolt István, David Sidler, Gustavo Alonso, Marko Vukolic * Systems Group, Department of Computer Science, ETH Zurich * Consensus IBM in Research,

More information

Conceptual Modeling on Tencent s Distributed Database Systems. Pan Anqun, Wang Xiaoyu, Li Haixiang Tencent Inc.

Conceptual Modeling on Tencent s Distributed Database Systems. Pan Anqun, Wang Xiaoyu, Li Haixiang Tencent Inc. Conceptual Modeling on Tencent s Distributed Database Systems Pan Anqun, Wang Xiaoyu, Li Haixiang Tencent Inc. Outline Introduction System overview of TDSQL Conceptual Modeling on TDSQL Applications Conclusion

More information

Staggeringly Large File Systems. Presented by Haoyan Geng

Staggeringly Large File Systems. Presented by Haoyan Geng Staggeringly Large File Systems Presented by Haoyan Geng Large-scale File Systems How Large? Google s file system in 2009 (Jeff Dean, LADIS 09) - 200+ clusters - Thousands of machines per cluster - Pools

More information

Using RDMA Efficiently for Key-Value Services

Using RDMA Efficiently for Key-Value Services Using RDMA Efficiently for Key-Value Services Anuj Kalia, Michael Kaminsky, David G. Andersen Carnegie Mellon University, Intel Labs CMU-PDL-14-16 June 214 Parallel Data Laboratory Carnegie Mellon University

More information

LS-DYNA Best-Practices: Networking, MPI and Parallel File System Effect on LS-DYNA Performance

LS-DYNA Best-Practices: Networking, MPI and Parallel File System Effect on LS-DYNA Performance 11 th International LS-DYNA Users Conference Computing Technology LS-DYNA Best-Practices: Networking, MPI and Parallel File System Effect on LS-DYNA Performance Gilad Shainer 1, Tong Liu 2, Jeff Layton

More information

Many-Core Computing Era and New Challenges. Nikos Hardavellas, EECS

Many-Core Computing Era and New Challenges. Nikos Hardavellas, EECS Many-Core Computing Era and New Challenges Nikos Hardavellas, EECS Moore s Law Is Alive And Well 90nm 90nm transistor (Intel, 2005) Swine Flu A/H1N1 (CDC) 65nm 2007 45nm 2010 32nm 2013 22nm 2016 16nm 2019

More information

CS 61C: Great Ideas in Computer Architecture. MapReduce

CS 61C: Great Ideas in Computer Architecture. MapReduce CS 61C: Great Ideas in Computer Architecture MapReduce Guest Lecturer: Justin Hsia 3/06/2013 Spring 2013 Lecture #18 1 Review of Last Lecture Performance latency and throughput Warehouse Scale Computing

More information

Reconfigurable hardware for big data. Gustavo Alonso Systems Group Department of Computer Science ETH Zurich, Switzerland

Reconfigurable hardware for big data. Gustavo Alonso Systems Group Department of Computer Science ETH Zurich, Switzerland Reconfigurable hardware for big data Gustavo Alonso Systems Group Department of Computer Science ETH Zurich, Switzerland www.systems.ethz.ch Systems Group 7 faculty ~40 PhD ~8 postdocs Researching all

More information

RoGUE: RDMA over Generic Unconverged Ethernet

RoGUE: RDMA over Generic Unconverged Ethernet RoGUE: RDMA over Generic Unconverged Ethernet Yanfang Le with Brent Stephens, Arjun Singhvi, Aditya Akella, Mike Swift RDMA Overview RDMA USER KERNEL Zero Copy Application Application Buffer Buffer HARWARE

More information

Tools for Social Networking Infrastructures

Tools for Social Networking Infrastructures Tools for Social Networking Infrastructures 1 Cassandra - a decentralised structured storage system Problem : Facebook Inbox Search hundreds of millions of users distributed infrastructure inbox changes

More information

Portland State University ECE 588/688. Directory-Based Cache Coherence Protocols

Portland State University ECE 588/688. Directory-Based Cache Coherence Protocols Portland State University ECE 588/688 Directory-Based Cache Coherence Protocols Copyright by Alaa Alameldeen and Haitham Akkary 2018 Why Directory Protocols? Snooping-based protocols may not scale All

More information

Baidu s Best Practice with Low Latency Networks

Baidu s Best Practice with Low Latency Networks Baidu s Best Practice with Low Latency Networks Feng Gao IEEE 802 IC NEND Orlando, FL November 2017 Presented by Huawei Low Latency Network Solutions 01 1. Background Introduction 2. Network Latency Analysis

More information

The Google File System

The Google File System The Google File System Sanjay Ghemawat, Howard Gobioff and Shun Tak Leung Google* Shivesh Kumar Sharma fl4164@wayne.edu Fall 2015 004395771 Overview Google file system is a scalable distributed file system

More information

Advanced Computer Networks. End Host Optimization

Advanced Computer Networks. End Host Optimization Oriana Riva, Department of Computer Science ETH Zürich 263 3501 00 End Host Optimization Patrick Stuedi Spring Semester 2017 1 Today End-host optimizations: NUMA-aware networking Kernel-bypass Remote Direct

More information

Facebook Tao Distributed Data Store for the Social Graph

Facebook Tao Distributed Data Store for the Social Graph L. Lancia, G. Salillari Cloud Computing Master Degree in Data Science Sapienza Università di Roma Facebook Tao Distributed Data Store for the Social Graph L. Lancia & G. Salillari 1 / 40 Table of Contents

More information

Size-aware Sharding For Improving Tail Latencies in In-memory Key-value Stores

Size-aware Sharding For Improving Tail Latencies in In-memory Key-value Stores Size-aware Sharding For Improving Tail Latencies in In-memory Key-value Stores Diego Didona EPFL Willy Zwaenepoel EPFL and University of Sydney Abstract This paper introduces the concept of size-aware

More information