High Bandwidth Data Acquisition and Network Streaming in VLBI Jan Wagner, Guifré Molera et al. TKK / Metsähovi Radio Observatory.

Slides:



Advertisements
Similar presentations
IBOB Data Acquisition and Network Streaming Jan Wagner et al. TKK / Metsähovi Radio Observatory 7th International eVLBI Workshop Shanghai, 2008.
Advertisements

SCARIe: Realtime software correlation Nico Kruithof, Damien Marchal.
Networking at Home the Office and Globe
Progress With iBOBs at Jodrell Bits & Bytes Meeting, JBO, th Dec 2007 Jonathan Hargreaves Electronic Engineer, Jodrell Bank Observatory.
Project Partners Project Collaborators The Council for the Central Laboratory of the Research Councils Funded by EPSRC GR/T04465/01
The CVN Real Time Correlator Zhang Xiuzhong, Chen Zhong Shanghai Astronomical Observatory China 4th e-VLBI Workshop Sydney
RAID- Redundant Array of Inexpensive Drives. Purpose Provide faster data access and larger storage Provide data redundancy.
Merit Network: Connecting People and Organizations Since 1966 CALEA Compliance – A Feasibility Study October 25, 2006 Mary Eileen McLaughlin Director –
Review on Networking Technologies Linda Wu (CMPT )
Using FPGAs with Embedded Processors for Complete Hardware and Software Systems Jonah Weber May 2, 2006.
Virtual Network Servers. What is a Server? 1. A software application that provides a specific one or more services to other computers  Example: Apache.
Ongoing e-VLBI Developments with K5 VLBI System Hiroshi Takeuchi, Tetsuro Kondo, Yasuhiro Koyama, and Moritaka Kimura Kashima Space Research Center/NICT.
GigE Knowledge. BODE, Company Profile Page: 2 Table of contents  GigE Benefits  Network Card and Jumbo Frames  Camera - IP address obtainment  Multi.
1 © 2005 Cisco Systems, Inc. All rights reserved. Cisco Public IP Telephony Introduction to VoIP Cisco Networking Academy Program.
VLBI Report LBA Operations & eVLBI Tasso Tzioumis ATNF May 2008.
Sven Ubik, Petr Žejdl CESNET TNC2008, Brugges, 19 May 2008 Passive monitoring of 10 Gb/s lines with PC hardware.
31st July 2008AIDA FEE Report1 AIDA Front end electronics Report July 2008 Progress Virtex5 FPGA choice Milestones for prototype delivery.
Performance Tradeoffs for Static Allocation of Zero-Copy Buffers Pål Halvorsen, Espen Jorde, Karl-André Skevik, Vera Goebel, and Thomas Plagemann Institute.
A+ Guide to Hardware: Managing, Maintaining, and Troubleshooting, Sixth Edition Chapter 9, Part 11 Satisfying Customer Needs.
5 Feb 2002Alternative Ideas for the CALICE Backend System 1 Alternative Ideas for the CALICE Back-End System Matthew Warren and Gordon Crone University.
Current LBA Developments Chris Phillips CSIRO ATNF 13/7/2005.
1 Computing Basics Honolulu Community College Cisco Academy Training Center Semester 1 Version
Towards a Common Communication Infrastructure for Clusters and Grids Darius Buntinas Argonne National Laboratory.
SEISLOG Linux presented at the WORKSHOP High Quality Seismic Stations and Networks for Small Budgets Volcan, Panama March, 2004 by Terje Utheim,
Cluster Computers. Introduction Cluster computing –Standard PCs or workstations connected by a fast network –Good price/performance ratio –Exploit existing.
Final Year Project Presentation by Daire O’Neill 4EE.
1 Next Few Classes Networking basics Protection & Security.
1 Selecting LAN server (Week 3, Monday 9/8/2003) © Abdou Illia, Fall 2003.
E-VLBI over TransPAC Masaki HirabaruDavid LapsleyYasuhiro KoyamaAlan Whitney Communications Research Laboratory, Japan MIT Haystack Observatory, USA Communications.
Understanding Data Acquisition System for N- XYTER.
Network Tests at CHEP K. Kwon, D. Han, K. Cho, J.S. Suh, D. Son Center for High Energy Physics, KNU, Korea H. Park Supercomputing Center, KISTI, Korea.
EVN-NREN meeting, Schiphol, , A. Szomoru, JIVE Recent eVLBI developments at JIVE Arpad Szomoru Joint Institute for VLBI in Europe.
Data transfer over the wide area network with a large round trip time H. Matsunaga, T. Isobe, T. Mashimo, H. Sakamoto, I. Ueda International Center for.
Chapter 2 Chapter 2: Planning for Server Hardware.
1 Lecture 7 LAN Wiring, Physical Topology, and Interface Hardware Computer Networks CS 4316.
VLBI Storage Manager (VSM) Matteo Stagni – Mauro Nanni 0.
Nov 3, 2009 RN - 1 Jet Propulsion Laboratory California Institute of Technology Current Developments for VLBI Data Acquisition Equipment at JPL Robert.
Chapter 1 Communication Networks and Services Network Architecture and Services.
VTP: VDIF Transport Protocol Chris Phillips, Alan Whitney, Mamoru Sekido & Mark Kettenis November 2011.
Introduction VLBI_UDP is an application being developed to transfer VLBI data using the UDP protocol. Initially developed by Richard Hughes- Jones for.
An FX software correlator for VLBI Adam Deller Swinburne University Australia Telescope National Facility (ATNF)
1 Presented By: Eyal Enav and Tal Rath Eyal Enav and Tal Rath Supervisor: Mike Sumszyk Mike Sumszyk.
High Speed Detectors at Diamond Nick Rees. A few words about HDF5 PSI and Dectris held a workshop in May 2012 which identified issues with HDF5: –HDF5.
Short introduction Pulsar Parkes. Outline PDFB – Single beam pulsar timing system CASPER – Single beam pulsar coherent dedispersion system.
VLBI_UDP An application for transferring VLBI data via UDP protocol Simon Casey e-VLBI meeting, Haystack 20 Sep 2006.
COMP381 by M. Hamdi 1 Clusters: Networks of WS/PC.
1 Masaki Hirabaru and Yasuhiro Koyama PFLDnet 2006 Febrary 2, 2006 International e-VLBI Experience.
1 eVLBI Developments at Jodrell Bank Observatory Ralph Spencer, Richard Hughes- Jones, Simon Casey, Paul Burgess, The University of Manchester.
Cluster Computers. Introduction Cluster computing –Standard PCs or workstations connected by a fast network –Good price/performance ratio –Exploit existing.
Introduction to Quality of Service Klara Nahrstedt CS 538.
Data delayed Real Time Data Transfer for Very Long Baseline Interferometry Simon Casey, Richard Hughes-Jones, Stephen Kershaw, Ralph Spencer, Matt Strong.
INDIANAUNIVERSITYINDIANAUNIVERSITY Tsunami File Transfer Protocol Presentation by ANML January 2003.
Studies of LHCb Trigger Readout Network Design Karol Hennessy University College Dublin Karol Hennessy University College Dublin.
NEXPReS Period 3 Overview WP8 FlexBuff High-Bandwidth, High-Capacity Networked Storage Ari Mujunen Aalto University Metsähovi Radio Observatory Finland.
COMPUTER NETWORKS CS610 Lecture-11 Hammad Khalid Khan.
WP8: FlexBuff Ari Mujunen, Tomi Salminen Aalto University Metsähovi Radio Observatory NEXPReS 2nd Board Meeting – 2012 June 20 – NORDUnet, Copenhagen Denmark.
Broadband Interfacing eMerlin-JIVE (1-10Gbps) Onsala-Jodrell Bank (
FABRIC WP1.2 Broadband Data Path: Protocols and Processor Interface Bonn 20/09/07 Ralph Spencer The University of Manchester.
ROD Activities at Dresden Andreas Glatte, Andreas Meyer, Andy Kielburg-Jeka, Arno Straessner LAr Electronics Upgrade Meeting – LAr Week September 2009.
G. Russo, D. Del Prete, S. Pardi Frascati, 2011 april 4th-7th The Naples' testbed for the SuperB computing model: first tests G. Russo, D. Del Prete, S.
Discussion for Compatibility. Common Recognition Data transport protocol in e-VLBI is ONLY important for realtime e-vlbi. Data (FILE) format is important.
Title FABRIC Progress - Mark, Huib etc. T. Charles Yun Project Manager EXPReS/JIVE.
Fermilab Scientific Computing Division Fermi National Accelerator Laboratory, Batavia, Illinois, USA. Off-the-Shelf Hardware and Software DAQ Performance.
Network customization
Introduction to Networks
Large CMS GEM with APV & SRS electronics
Storage Virtualization
VTP: VDIF Transport Protocol
iBOB Data Acquisition and Network Streaming
Network customization
Presentation transcript:

High Bandwidth Data Acquisition and Network Streaming in VLBI Jan Wagner, Guifré Molera et al. TKK / Metsähovi Radio Observatory

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #2 EXPReS: funds to connect stations at 1G/10G 1G fiber up to capacity with Tsunami UDP (2006) or JIVE vlbi_udp (2007) real-time transport protocols ToDo: new system for >1 Gbps rates! Basic e-VLBI Data Acquisition Introduction Basic: 8–16 stations, rates 128–512 Mbit/s; 24h 1–4 TB/station, all to correlator VLBI: observed galaxy’s noise recorded to disks, shipped to correlator e-VLBI: transfer station data via fast research networks, two alternatives: post-observation: record locally, transfer to correlator later real-time: stream to the correlator for temp storage or processing 1Apr08, JIVE vlbi_udp incoming aggregate rate record

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #3 Goals for High(er) Bandwidth DAQ Observe a broader radio spectrum broader spectrum => higher data rate correlation result and final image improves! high EXPReS goal 1/2/4/8 Gbps per station Higher rates => need at least: new analog-to-digital converter system (current 10+ years old) faster, higher capacity local/remote data storage systems suitable transport protocols - stations to correlator Ultimately e-VLBI data processing in real-time get rid of high capacity station data storage JIVE: hardware correlator successful in many ≤ 512 Mbps real-time runs distributed software correlators more natural with data streaming Introduction C. Phillips, e-VLBI Scientific Benefits, ATNF CSIRO memo, 2004

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #4 High Bandwidth DAQ Development DAQ steps: Digitize wide-band signal – Stream over 10G – Capture/store with PC(s) Development Environment

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #5 DAQ – Digitizing UC Berkeley’s iBOB FPGA hardware platform Virtex2 Pro FPGA with PowerPC405 two CX4 10G ports 1..2 Berkeley iADC expansion boards Atmel AT84AD001 two channels, 8-bit, 1 Gsample/s => max 8+8 Gbps from one iADC => 32 Gbps ADC data, 20 Gbps net Semi-ready component libraries from Berkeley for “faster” start of development Analog signal digitizing setup: iBOB FPGA reads raw 2-channel iADC data FPGA processes data, implements network streaming => … Basic Data Acquisition System

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #6 DAQ – Streaming TCP is not a good streaming choice No unreliable best-effort delivery, no delivery time constraints Real-time VLBI stream requirements: data loss f≤10% is tolerable (impact is ), occasional re-order ok data integrity non-critical: BER*2·10 -9 frame CRC32, checksumless UDP data must have time stamp: a 64-bit PSN sufficient, congestion control: none, data thinning, change sampling mode iBOB UDP streaming first implementation iBOB has only 14MB RAM Gbps) => no retransmissions => plain UDP packetizer, 1 stream per channel streaming configurable remotely via Telnet/ssh overall iBOB test design… =>. Basic Data Acquisition System

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #7 DAQ – iBOB RF-to-10G Test Design Basic Data Acquisition System

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #8 DAQ – Storage / RAID Performance Should use RAID-0 disk striping for best throughput Hardware RAID e.g. PCIe cards, XRaid, VTrak,.. too expensive and slow Software RAID baseline performance test on: ~1k€ workstation PC, Linux RAID-0 12 x SpinPoint F1, 9TB Test method: - source buffer ~90% RAM size - sequential fwrite() to a single file - not using 10G / UDP yet => Inexpensive 1–4 Gbps data storage system! Basic Data Acquisition System

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #9 DAQ – Stream Recording Wrote simple UDP/raw capture programs udp2raid, raw2raid Essential for performance in 10G stream recording: 10G NIC throughput: 4.7/9K jumbo frames, UDP checksums off prefer larger UDP packets, larger OS buffers => Myri, Chelsio NIC PC iperf : 9.9 Gbps UDP, 0% loss Application and disk throughput: use multi-core system, try to utilize (nearly) all cores  separate network and disk I/O into own threads => udp2raid, raw2raid currently min. 4 Gbps from 10G to RAID-0 => we have 1–4 Gbps iBOB stream to Abidal PC disk recording Basic Data Acquisition System

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #10 Fast “off-line” e-VLBI File Transfer Transfer station files to where you process the data, one option: Tsunami UDP v1.1 fast ‘FTP’ and real-time ( Weekly used in geodesic VLBI for ≤1 Gbps file transfer, occasional real-time Current v1.1 version, first result on 10G LAN: high CPU load, 3 – 3.5 Gbps to RAID, protocol rate limit 4 Gbps Started v1.2 of protocol to use udp2raid improvements: 4 – 4.4 Gbps to Abidal PC RAID 7 Gbps to RAM (vs. UDT4-protocol 5.2 Gbps to RAM) Basic Data Acquisition System => Improved rates for “off-line” e-VLBI

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #11 Summary Many European VLBI stations start to reach ~1 Gbps streaming with the old hardware Basic iBOB design can further improve the rate: digitize and stream 1 – 20 Gbps UDP A consumer ~1 k€ PC + 12 disks is capable of 4 Gbps storage: can capture the iBOB stream(s) to PC RAID Tsunami v1.2 has improved file transfer rates Can transfer from station to similar ≥4 Gbps PC elsewhere ToDo: port real-time Tsunami server onto iBOB board

10 April 2008EXPReS- High Bandwidth Data Acquisition and Network Streaming in VLBISlide #12 Questions/Answers Contact information Jan Wagner M.Sc./Researcher EXPReS (Metsähovi) Additional Information [note: only one “s”] EXPReS is made possible through the support of the European Commission (DG-INFSO), Sixth Framework Programme, Contract #026642