Data Center Networks for the Application

Slides:



Advertisements
Similar presentations
Fast Data at Massive Scale Lessons Learned at Facebook Bobby Johnson.
Advertisements

Big Data + SDN SDN Abstractions. The Story Thus Far Different types of traffic in clusters Background Traffic – Bulk transfers – Control messages Active.
Logically Centralized Control Class 2. Types of Networks ISP Networks – Entity only owns the switches – Throughput: 100GB-10TB – Heterogeneous devices:
MapReduce.
DOT – Distributed OpenFlow Testbed
Algorithms and Problem Solving-1 Algorithms and Problem Solving.
Design Patterns for Efficient Graph Algorithms in MapReduce Jimmy Lin and Michael Schatz University of Maryland Tuesday, June 29, 2010 This work is licensed.
Introduction to Boosting Aristotelis Tsirigos SCLT seminar - NYU Computer Science.
Information-Agnostic Flow Scheduling for Commodity Data Centers
SM3121 Software Technology Mark Green School of Creative Media.
Practical TDMA for Datacenter Ethernet
Advanced Topics: MapReduce ECE 454 Computer Systems Programming Topics: Reductions Implemented in Distributed Frameworks Distributed Key-Value Stores Hadoop.
Cloud Computing April 19, Agenda Introduction What is cloud computing? What are the benefits? What are the concerns? What’s next?
Assumptions Hypothesis Hopes RAMCloud Mendel Rosenblum Stanford University.
Jon Turner (and a cast of thousands) Washington University Design of a High Performance Active Router Active Nets PI Meeting - 12/01.
Jon Crowcroft, Data Center Networks for the Application.
David G. Andersen CMU Guohui Wang, T. S. Eugene Ng Rice Michael Kaminsky, Dina Papagiannaki, Michael A. Kozuch, Michael Ryan Intel Labs Pittsburgh 1 c-Through:
CS 395 Last Lecture Summary, Anti-summary, and Final Thoughts.
Wei Bai with Li Chen, Kai Chen, Dongsu Han, Chen Tian, Hao Wang SING HKUST Information-Agnostic Flow Scheduling for Commodity Data Centers 1 SJTU,
Virtual Machines Created within the Virtualization layer, such as a hypervisor Shares the physical computer's CPU, hard disk, memory, and network interfaces.
An Efficient Gigabit Ethernet Switch Model for Large-Scale Simulation Dong (Kevin) Jin.
Interconnect Networks Basics. Generic parallel/distributed system architecture On-chip interconnects (manycore processor) Off-chip interconnects (clusters.
An Efficient Gigabit Ethernet Switch Model for Large-Scale Simulation Dong (Kevin) Jin.
Data Centers and Cloud Computing 1. 2 Data Centers 3.
Three topics – 1. Data Centers, 2. IoT, 3. Networks for Dev/Disasters Jon Crowcroft,
The Post Windows Operating System
Complexity Analysis (Part I)
Big Data is a Big Deal!.
Algorithms and Problem Solving
CPU SCHEDULING.
Slicer: Auto-Sharding for Datacenter Applications
An Open Source Project Commonly Used for Processing Big Data Sets
Analysis of Algorithms
Cloud Legal Project - Update
Shadow: Real Applications, Simulated Networks
Haim Kaplan and Uri Zwick
Ch. 11 Theory of Computation
Software Architecture in Practice
Srinivas Narayana MIT CSAIL October 7, 2016
Plethora: Infrastructure and System Design
The Stanford Clean Slate Program
湖南大学-信息科学与工程学院-计算机与科学系
Chapter 14: Protection.
James B. Orlin Presented by Tal Kaminker
CPU Scheduling Basic Concepts Scheduling Criteria
CPU Scheduling G.Anuradha
Cse 344 May 4th – Map/Reduce.
Shadow: Scalable and Deterministic Network Experimentation
Reliable Distributed Systems
Distributed System Gang Wu Spring,2018.
Outline Module 1 and 2 dealt with processes, scheduling and synchronization Next two modules will deal with memory and storage Processes require data to.
Software models - Software Architecture Design Patterns
Big-O, Ω, ϴ Trevor Brown CSC263 Tutorial 1 Big-O, Ω, ϴ Trevor Brown
Fast Congestion Control in RDMA-Based Datacenter Networks
CSE 311: Foundations of Computing
Algorithms and Problem Solving
Priority Queues An abstract data type (ADT) Similar to a queue
Network Processors for a 1 MHz Trigger-DAQ System
Centralized Arbitration for Data Centers
Introduction to MapReduce
The Challenge of Cross - Language Interoperability
CS 6290 Many-core & Interconnect
Recitation #4 Tel Aviv University 2017/2018 Slava Novgorodov
CIS679: Two Planes and Int-Serv Model
Your Programmable NIC Should Be a Programmable Switch
Complexity Analysis (Part I)
Jennifer Rexford Princeton University
Presentation transcript:

Data Center Networks for the Application Jon Crowcroft, http://www.cl.cam.ac.uk/~jac22

Data Centers don’t just go fast They need to serve applications Latency, not just throughput Availability & Failure Detectors Application code within network Work based on Matt Grosvenor’s PhD And EPSRC funded NaaS project…

1. Deterministic latency bounding Learned what I was teaching wrong! I used to say: Integrated Service too complex Admission&scheduling hard Priority Queue can’t do it PGPS computation for latency? I present Qjump scheme, which Uses intserv (PGPS) style admission ctl Uses priority queues for service levels

Data Center Latency Problem Tail of the distribution, due to long/bursty flows interfering Need to separate classes of flow Low latency are usually short flows (or RPCs) Bulk transfers aren’t so latency/jitter sensitiv

Data Center Qjump Solution In Data Center, not general Internet! can exploit topology & traffic matrix & source behaviour knowledge Regular, and simpler topology key But also largely “cooperative” world…

5 port virtual output queued

Latency of ping v. iperf

Hadoop perturbs time synch

Hadoop perturbs memcached

Hadoop perturbs Naiad

Qjump – two pieces At network config time At run time Compute a set of (8*) rates based on Traffic matric & hops => fan in (f) At run time Flow assigns itself a priority/rate class subject it to (per hypervisor) rate limit * 8 arbitrary – but often h/w supported

Qjump – self admit/policer (per hypervisor, at least)

Qjump latency thru 1 switch

Memcached latency redux w/ QJ

QJ naiad barrier synch latency redux

Experimental topology

PTPd/memcached along, with Hadoop, with Hadoop+QJump

2PC throughput with and without QJump

QJump: variance compared with other schemes…

Topology for flow completion experiment

Web search FCT100Kb ave

Web search FCT 99th centile

Web search FCT 10Mb ave

Data-mining 100k FCT ave

Data-mining 100k FCT 99th cent

Data-mining 10M FCT ave

Memcached latency

Qjump scale-up emulation

Latency bound validation topo

Big Picture Comparison – Related work…

2 Failure Detectors 2PC & CAP theorem Recall CAP (Brewer’s Hypothesis) Consistency, Availability, Partitions Strong& weak versions! If have net&node deterministic failure detector, isn’t necessarily so! What can we use CAP-able system for?

Consistent, partition tolerant app? Software Defined Net update! Distributed controllers have distributed rules Rules change from time to time Need to update, consistently Need update to work in presence of partitions By definition! So Qjump may let us do this too!

3. Application code -> Network Last piece of data center working for application Switch and Host NICs have a lot of smarts Network processors, like GPUs or (net)FPGAs Can they help applications? In particular, avoid pathological traffic patterns (e.g. TCP incast)

Application code E.g. shuffle phase in map/reduce Code very simple Does a bunch of aggregation (min, max, ave) on a row of results And is cause of traffic “implosion” So do work in stages in the switches in the net (like merge sort!) Code very simple Cross-compile into switch NIC cpus

Other application examples Are many … Arose in Active Network research Transcoding Encryption Compression Index/Search Etc etc

Need language to express these Finite iteration (not Turing-complete language) So design python– with strong types! Work in progress in NaaS project at Imperial and Cambridge…

Conclusions/Discussion Data Center is a special case! Its important enough to tackle We can hard bound latency easily We can detect failures and therefore solve some nice distributed consensus problems We can optimise applicatiosn pathological traffic patterns Plenty more to do…