NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
The Present Connectivity
- NRAO Intranet
- Frame relay
- Fractional T1 at all four of NRAO's major sites
- VLA and VLBA PT connected via tie line to the AOC
- 12 Meter and VLBA KP connected through NOAO and UofA
- Low speed connection at 6 VLBA sites
- VLBA LA and MK not connected
- Also connection to NCSA for AIPS++ collaboration
- Internet connectivity
- University of Virginia (CV & GB)
- New Mexico Institute of Mining and Technology (SOC)
- University of Arizona (TUC)
- Connections to NSF vBNS
- ``Back-door'' connections to vBNS at UVa, UofA
- Legitimate connection at NCSA
Top
NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
vBNS & Internet2
- vBNS
- MCI Worldcom is present contractor
- Approx 150 institutions connected (all? are major universities)
- Subscription charge paid by each site connecting
- $45k/yr for OC-3 (155 Mbps)
- $135k/yr for OC-12 (620 Mbps)
- Can only share cost at a GigaPoP if >5 connections
- Internet2
- aka ``Abilene''
- ``Under development by the University Corporation for Advanced Internet
Development (UCAID) in partnership with Qwest Communications, Nortel
(Northern Telecom) and Cisco Systems.''
- ``Scheduled for initial operation in early 1999''
- Membership $25k/yr
- Abilene participation fee $20k/yr per connecting site
- Abilene connection fee for OC-3: $110k/year per connection
- Abilene connection fee for OC-12: $320k/year per connection
- Connection fees are shared at a GigaPoP
- Notes
- These costs do not include line charges
- NRAO is considering DS-3 (45 Mbps) connections
- The connection protocol is ATM
Top
NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
NSF Astronomy Division Initiative
- Meeting at NSF on January 26, 1999
- NRAO, NOAO, Gemini
- Not NAIC, because only intra-USA
- NSF wanted proposal by mid-March (initially)
- Not enough time for the proposal, so did three instead
- A single grant is for a maximum of $350k (plus subscription or
participation fees)
- Mainly presented the case for VLA, but neither VLA nor GBT ready to
produce high speed real-time data
- Bad timing
- On February 12, we were told to go ahead with a proposal for the
initiative
- Initial focus on VLA, VLBA
- Discovered UNM, NMIMT, SIPI will propose GigaPoP at UNM to connect major
labs
- Meeting March 02 with NMIMT
- Zero cost to add AOC
- Can add SOC-VLA for ~$90k/yr in the future
- Gustaaf van Moorsel is now tracking this
- Tony Beasley had drafted an abstract proposal in summer 1998 for enhanced
inter-connectivity and video conferencing
- Asked on March 08 to submit proposal by March 10 !!
- Funded March 17
- ``We'll have to be careful what we ask for next time, we might get it
...''
- Concentrated on GB for the response to the initiative
- April 01 we were ready to go, but ...
Top
NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
NRAO Intranet Upgrade and Videoconferencing
- All four major NRAO sites will have T1 access
- T1 PVCs between the sites (probably)
- Teleconferencing equipment also at the four sites
- (Where do we put it - auditorium or conference room?)
- ATM switch to be able to handle the multiple teleconferencing sites
- (Could also be used to connect to vBNS, Internet2. Funds can also be
redirected.)
- Fiber connection from 12 Meter to kitchen on KP
- 1 FTE for real-time remote observing applications
- $450k total
- Funds expected in June
- Phil Jewell is PI
- Time Cornwell, Gareth Hunt, CoIs
- Alan Bridle, Gene Runion, and Tony Beasley contributed enormously
Top
NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
Major Needs Requiring High Speed Networking
- Interactive remote use of NRAO telescopes
- On-line access to archived telescope data
- High-performance imaging
- Enhanced conferencing, between NRAO and user institutions and between NRAO
sites
- Public outreach
Top
NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
Green Bank Proposal (1)
- State-of-the-art Telescope
- National Radio Quiet Zone
- Other instruments in Green Bank - GBES, GBI, USNO VLBI
- GB physically remote
- Dynamic scheduling for weather and interference
- Dynamic scheduling requires good communications
- ``Just like being there''
- Services just as being used at the 12 Meter
- Audio/Visual services
- Weather data, telescope status, monitor data, etc.
- Interaction with operator
- Single dish observing traditionally more interactive than synthesis arrays
- High data volume observing
- Spectrometer
- On-The-Fly Mapping
- (pulsar)
- High bandwidth
- 25 Mbps for 10 Hz video
- At 3mm, 10x10 arcminutes (176x176 pixels), 1024 channels, 4 bytes = 1
Gbit
- Observed in ~180 seconds implies a maximum needed data rate of ~6 Mbps
Top
NRAO Real-time Programmers Meeting 12 April 1999
High Speed Networking at the NRAO
Green Bank Proposal (2)
- M&C already use modern network protocols (RPCs, etc.)
- For software running at remote institutions we need multiple architecture
support
- AIPS++ already does this (glish therefore also)
- M&C has very limited platform independence (except glish)
- Remote X displays do not have this restriction, but are less efficient for
communications
- Network support issues
- Security
- Tool to permit remote access to authorized users
- Need this for operation within Green Bank anyway
- Intermittent connections
- Variable speed and latency
- Pricing
- GB is 90 miles from any PoP (Charleston, WV) - $10k/mon access
- Rest of service is ~$9k/mon
- Irrespective of terminal location
- Blacksburg, Va (VaTech)
- Bluefield, WV
- Charlottesville, Va (NRAO, UVa)
- Morgantown, WV (WVU)
- Independent of protocol or line
- Bottom line cost is therefore $350k per year
- Grant would cover only one year of operation