Data Path through host/ANP.

Slides:



Advertisements
Similar presentations
Distributed Processing, Client/Server and Clusters
Advertisements

System Area Network Abhiram Shandilya 12/06/01. Overview Introduction to System Area Networks SAN Design and Examples SAN Applications.
Fast Communication Firefly RPC Lightweight RPC  CS 614  Tuesday March 13, 2001  Jeff Hoy.
Embedded Web Hyung-min Koo. 2 Table of Contents Introduction of Embedded Web Introduction of Embedded Web Advantages of Embedded Web Advantages of Embedded.
Cache Coherent Distributed Shared Memory. Motivations Small processor count –SMP machines –Single shared memory with multiple processors interconnected.
Panoptes: A Scalable Architecture for Video Sensor Networking Applications Wu-chi Feng, Brian Code, Ed Kaiser, Mike Shea, Wu-chang Feng (OGI: The Oregon.
1 In VINI Veritas: Realistic and Controlled Network Experimentation Jennifer Rexford with Andy Bavier, Nick Feamster, Mark Huang, and Larry Peterson
The Virtual Microscope Umit V. Catalyurek Department of Biomedical Informatics Division of Data Intensive and Grid Computing.
Modifying the SCSI / Fibre Channel Block Size Presented by Keith Bonneau, John Chrzanowski and Craig O’Brien Advised by Robert Kinicki and Mark Claypool.
Active Messages: a Mechanism for Integrated Communication and Computation von Eicken et. al. Brian Kazian CS258 Spring 2008.
Virtualization for Cloud Computing
Understanding and Managing WebSphere V5
Fundamentals of Computer Networks ECE 478/578 Lecture #2 Instructor: Loukas Lazos Dept of Electrical and Computer Engineering University of Arizona.
LIGO-G E ITR 2003 DMT Sub-Project John G. Zweizig LIGO/Caltech Argonne, May 10, 2004.
©Ian Sommerville 2006Software Engineering, 8th edition. Chapter 12 Slide 1 Distributed Systems Architectures.
Chapter 6 Operating System Support. This chapter describes how middleware is supported by the operating system facilities at the nodes of a distributed.
DISTRIBUTED COMPUTING
Extracted directly from:
Eric Keller, Evan Green Princeton University PRESTO /22/08 Virtualizing the Data Plane Through Source Code Merging.
Dynamic Reconfiguration Dynamic selection of handler functionality: currently through use of parameterizable handlers or by selecting from a set of existing.
Uncovering the Multicore Processor Bottlenecks Server Design Summit Shay Gal-On Director of Technology, EEMBC.
BLU-ICE and the Distributed Control System Constraints for Software Development Strategies Timothy M. McPhillips Stanford Synchrotron Radiation Laboratory.
G-Number 1 Forwarding and Control Element Separation (ForCES) Overview & Requirements Update Todd A. Anderson.
LIGO-G9900XX-00-M ITR 2003 DMT Sub-Project John G. Zweizig LIGO/Caltech.
Heavy and lightweight dynamic network services: challenges and experiments for designing intelligent solutions in evolvable next generation networks Laurent.
Stream Processing Main References: “Comparing Reyes and OpenGL on a Stream Architecture”, 2002 “Polygon Rendering on a Stream Architecture”, 2000 Department.
© 2004 Mercury Computer Systems, Inc. FPGAs & Software Components Graham Bardouleau & Jim Kulp Mercury Computer Systems, Inc. High Performance Embedded.
MAPLD Reconfigurable Computing Birds-of-a-Feather Programming Tools Jeffrey S. Vetter M. C. Smith, P. C. Roth O. O. Storaasli, S. R. Alam
Students: Anurag Anjaria, Charles Hansen, Jin Bai, Mai Kanchanabal Professors: Dr. Edward J. Delp, Dr. Yung-Hsiang Lu CAM 2 Continuous Analysis of Many.
Supporting Runtime Reconfiguration on Network Processors Kevin Lee Lancaster University
Tool Integration with Data and Computation Grid GWE - “Grid Wizard Enterprise”
1 Public DAFS Storage for High Performance Computing using MPI-I/O: Design and Experience Arkady Kanevsky & Peter Corbett Network Appliance Vijay Velusamy.
GVis: Grid-enabled Interactive Visualization State Key Laboratory. of CAD&CG Zhejiang University, Hangzhou
What is SAM-Grid? Job Handling Data Handling Monitoring and Information.
A Utility-based Approach to Scheduling Multimedia Streams in P2P Systems Fang Chen Computer Science Dept. University of California, Riverside
Harmony: A Run-Time for Managing Accelerators Sponsor: LogicBlox Inc. Gregory Diamos and Sudhakar Yalamanchili.
ND The research group on Networks & Distributed systems.
Intel Research & Development ETA: Experience with an IA processor as a Packet Processing Engine HP Labs Computer Systems Colloquium August 2003 Greg Regnier.
IXP1200 Applications Ada Gavrilovska, Jiantao Kong, Weidong Shi, Xiaotong Zhuang Dr. Karsten Schwan, Dr. Ken Mackenzie Scalable Real Time Media Streaming.
Investigating the Performance of Audio/Video Service Architecture I: Single Broker Ahmet Uyar & Geoffrey Fox Tuesday, May 17th, 2005 The 2005 International.
A SEMINAR ON 1 CONTENT 2  The Stream Programming Model  The Stream Programming Model-II  Advantage of Stream Processor  Imagine’s.
Full and Para Virtualization
An Overview of Scientific Workflows: Domains & Applications Laboratoire Lorrain de Recherche en Informatique et ses Applications Presented by Khaled Gaaloul.
High Performance Flexible DSP Infrastructure Based on MPI and VSIPL 7th Annual Workshop on High Performance Embedded Computing MIT Lincoln Laboratory
ECE 526 – Network Processing Systems Design Programming Model Chapter 21: D. E. Comer.
Software Systems Division (TEC-SW) ASSERT process & toolchain Maxime Perrotin, ESA.
Addressing Data Compatibility on Programmable Network Platforms Ada Gavrilovska, Karsten Schwan College of Computing Georgia Tech.
Danilo Florissi, Yechiam Yemini (YY), Sushil da Silva, Hao Huang Columbia University, New York, NY 10027
Software Architecture of Sensors. Hardware - Sensor Nodes Sensing: sensor --a transducer that converts a physical, chemical, or biological parameter into.
Distributed Processing, Client/Server and Clusters
Univa Grid Engine Makes Work Management Automatic and Efficient, Accelerates Deployment of Cloud Services with Power of Microsoft Azure MICROSOFT AZURE.
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING CLOUD COMPUTING
Clouds , Grids and Clusters
SuperComputing 2003 “The Great Academia / Industry Grid Debate” ?
Digital Map Server for Real Time Embedded Platforms
VirtualGL.
Grid Computing.
Adapting Applications and Platforms
Texas Instruments TDA2x and Vision SDK
University of Technology
Cloud Computing.
1. 2 VIRTUAL MACHINES By: Satya Prasanna Mallick Reg.No
OASIS Deployment: VideoCollective
Wavestore Integrates…
Virtualization Techniques
Inventory of Distributed Computing Concepts
Hadoop Technopoints.
Cloud computing mechanisms
AWS Cloud Computing Masaki.
An XML-based System Architecture for IXA/IA Intercommunication
Presentation transcript:

Data Path through host/ANP. Application-level Services on Attached Network Processors Ada Gavrilovska, Karsten Schwan, Austen Mcdonald, Hailemelekot Seifu, Ola Nordstrom Objectives Utilize programmable network processors in large scale distributed applications to: attain improvements in end-user application performance, more efficiently utilize host system capacity, offer to applications new services at no additional performance overheads perceived by end users. Map application- and middleware-level services across hosts and their attached NPs, thereby attaining integrated host-attached NP (ANP) systems SPLITS Overview Stream Handler - lightweight computational unit executed on the ANP which operates on both, packet header and payload data SPLITS – Software architecture for Programmable LIghtweighT Stream handlers enables developers to dynamically deploy and configure service functionality onto attached network processors so as to best utilize the combined host/ANP resources implemented for standard Linux kernels and ANPs based on the Intel IXP network processors. Receive Transmit Access user kernel protocol plane host ANP from network to network Data Path through host/ANP. Application- and middleware-level services are implemented as compositions of stream handlers simultaneously deployed at multiple points along the packets’ datapath through the integrated host/ANP nodes: on the ANP, embedded with packet receipt and transmission, on the ANP, after the application-level message has been assembled in ANP memory, or at the host kernel or user level Rx SH Tx Control Mgt Data Mgt Control Data Data Buffers resource state ANP-HOST INTERFACE HOST ANP Resource Monitor Admission Control Application h Applications: commercial (OIS), scientific (RT collaborations), dynamic web applications (delivery of camera-captured data; Services: application-level filtering, mirroring, stream differentiation, format translation, downsampling… Dynamic Reconfiguration Dynamic selection of handler functionality: currently through use of parameterizable handlers or by selecting from a set of existing handlers assessing possibilities for runtime deployment of new codes Admission control: monitor use of resources such as headroom consider both IXP- and host-side deployment Accessing application-level data: efficient protocol for fragmentation and reassembly of application-level messages (RUDP) use of portable binary data format descriptors (PBIO) Host - Attached NP node: integrated platforms, building block in overlay networks, interactive grids, p2p systems, cluster servers… Summary of Results: Variety of application-level services can be efficiently implemented Computationally intensive services such as image cropping for OpenGl graphics serves can be supported on the IXP NP at rates reaching 3.75Gbps; Value-added services, such as format translation for XML-based applications, can be added at no additional cost perceived by end user; Results indicate the importance of cooperative host/ANP processing for complex services where handlers are simultaneously deployed on the ANP and the host; Performance gains stem from network-near execution of stream handlers, load reduction on the host system CPU and memory infrastructure, and the flexibility with which stream handlers can be mapped across host/ANP boundaries sample services implemented with stream handlers and SPLITS data format translation on the IXP for XML-based applications