20011004-GigapopTransportOptions-Winkler

Download Report

Transcript 20011004-GigapopTransportOptions-Winkler

GigaPoP Transport Options:
I-WIRE Positioning for the Bandwidth
Tsunami
Virtual Internet2 Member Meeting
Oct 4, 2001
Linda Winkler
Argonne National Laboratory
[email protected]
I-WIRE Background
• State Funded Dark Fiber Optical
Infrastructure to support
Networking and Applications
Research
UIC
NU / Starlight
Star Tap
ANL
IIT
UC
• $7.5M Total Funding
• $4M FY00-01 (in hand)
• $2.5M FY02 (approved 1-June-01)
• Additional $0.5M in FY03, FY04
• Application Driven
NCSA/UIUC
• Access Grid: Telepresence & Media
• Computational Grids: Internet
Computing
• Data Grids: Information Analysis
• New Technologies Proving Ground
• Optical Switching
• Dense Wave Division Multiplexing
• Advanced middleware infrastructure
For more information see www.iwire.org
Dark Fiber
• Location, location, location
• Metro
• Roughly $2 per meter per strand
• Lateral challenges
• Regional
• May require regeneration; regen, reshape, retime (3R) is expensive!
• Wide area
• Expensive due to 3R requirements
• Key is location of carrier runs
• $100K-$1M/month for 2200 mile OC-192 link
Dark Fiber (cont.)
• Key is the one time up front cost for the purchase of an IRU
• Maintenance and management are the buyers problem
• Obtain fiber characteristics as soon as possible (SMF vs.
NZDSF, OTDR shots)
• Rapid provisioning possible allowing more adaptive
networks
• The fiber industry is immature and underdeveloped,
allowing sophisticated customers to negotiate much more
attractive deals than would result from a standard RFP
pricing exercise.
Indefeasible Right to Use (IRU) Services
• Terms 10, 15, 20 yrs
• Alternatives
• Long term capital lease
• Short term lease
• Managed service
• Considered as a physical asset which can be re-sold, traded
or used a collateral.
• Cost can be amortized over lifetime which results in a
monthly cost substantially below traditional
telecommunication services.
• Be sure of contract conditions due to shaky nature of some
vendors financial situation.
Dense Wavelength Division Multiplexing
• To ring or not to ring (ring vs. mesh)
• Redundancy (at what cost?)
• Survivability
• Protection
• Mesh topology benefits
•
•
•
•
•
Migration, scaling
Deployment speed
Capacity Utilization
Network Restoration
Operating Costs
• Laser reach-3R issues and OEO
• Large portion of the cost
• No standards
• Number and spacing of lambdas are design variables
• One transponder per wavelength
• Beware OC-192/10GbE WAN PHY/10GbE LAN PHY are not all equal
TeraGrid Backplane
StarLight
International Optical Peering Point
(see www.startap.net)
Abilene
Chicago
St. Louis
Indianapolis
Urbana
Los Angeles
San Diego
OC-48 (2.5 Gb/s, Abilene)
Multiple 10 GbE (Qwest)
Multiple 10 GbE (I-WIRE Dark Fiber)
• Solid lines in place and/or available by October 2001
• Dashed I-WIRE lines planned for Summer 2002
UIC
I-WIRE
Starlight / NW Univ
Multiple Carrier Hubs
ANL
Ill Inst of Tech
St Louis
GigaPoP
Univ of Chicago
Indianapolis
(Abilene NOC)
NCSA/UIUC
Charlie Catlett – Argonne National Laboratory ([email protected])
TeraGrid Proposed Backplane Architecture
One Wilshire
(Carrier Fiber Collocation Facility)
455 N. Cityfront Plaza
(Qwest Fiber Collocation Facility)
4 x 10 GbE
Los Angeles
Vendor TBD
Long-Haul DWDM
(Operated by site)
Chicago
2200mi
DTF
Backbone
Core Switch
Vendor TBD
Metro DWDM
(operated by site)
15mi
115mi
Vendor POP
at JPL
Qwest San
Diego POP
2mi
140mi
25mi
Vendor TBD
Metro DWDM
20mi
Caltech
SDSC
Ciena CoreStream™
Long-Haul DWDM
(Operated by Qwest)
ANL
NCSA
Vendor TBD
Switch/Router*
(256 Gb/s crossbar)
Site Border
Switch
Cluster
Aggregation
Switch
Caltech
Cluster
(64p)
SDSC
Cluster
(250p)
NCSA
Cluster
(2000p)
Charlie Catlett – Argonne National Laboratory ([email protected])
ANL
Cluster
(128p)
DTF Local Site
Resources and
External Network
Connections
*Initial Phase using IP Switch/Routers.
This design will be evaluated beginning October
2001. Phase 2 (optical mesh) will also be
evaluated prior to full DTF cluster deployment in
early 2002..
Wavelength Services
• OC48, OC192 vs. 10GbE
• Be sure of handoff specifications
• Management- determine level required
• Qwest/Teleglobe/(3)Link Global/Global Crossing service
offerings
• Benefits
•
•
•
•
•
•
•
Lower cost
Customer responsible for protection
Share cost of electronics-save capital investment
Customer empowerment
Potential for higher utilization of network
Transparency of signal
Flexibility
IP Routers and Switches
• Interoperability
• Which PHY (LAN vs. WAN) interface between DWDM and CPE?
• $$$
• WAN PHYs tend to be pricey
• Is 10GbE really 10,000 Mb/s, or 9.3 Gb/s, or maybe 8 Gb/s?
• What is the largest individual stream you must support?
• Aggregates vs. large streams
• Currently no visibility into the optical layer
Next Steps: Optical Mux / Wavelength Router /
Optical Wavelength Cross-connect System
2.5 / 10 / 40 Gb/s
GigE / 2.5 Gb/s 10 Gb/s
OTU
l1 DWDM
Optical Mux/
l-Router/
Cross-Connect
Customer
Interface
ln
OTU
FDP
lx
Optical Switches
• Current tech is O-E-O
•
•
•
•
•
Incoming signals are converted from Optical to Electrical
Signal is switched electrically
Outgoing signals are converted back to optical
Up to 64x64 at 2.5 Gb/s
Smart but slow
• Future tech will be O-O-O
•
•
•
•
•
•
Operate in the all optical domain
Bit rate independent
A win at 40 Gb/s
Below 10 Gb/s electronic switching will be hard to beat
Challenge will be in management
Fast but dumb
Optical Internetworking Progress
• Optical Internetworking Forum (OIF)
• UNI 1.0 specification in progress
• Based on domain services model
• Internet Engineering Task Force (IETF)
• Generalized Multiprotocol Label Switching (GMPLS)
• Incorporates Domain and Peer Models
• OIF and IETF are in sync
• OIF UNI 1.0 based on GMPLS
• Start with Domain Model and evolve
• Stay tuned. The ending of this story has not yet been
written.
Wavelengths and the Future
• Wavelength services are causing a network
revolution:
• Core long distance SONET Rings will be replaced by meshed
networks using wavelength cross-connects
• Re-invention of pre-SONET network architecture
• Improved transport infrastructure will exist for
IP/packet services
• Electrical/Optical grooming switches will emerge at
edges
• Automated Restoration (algorithm/GMPLS driven)
becomes technically feasible.
• Operational implementation will take beyond 2003
StarLight Optical Peering Point & Co-lo Facility
• 710 N. Lake Shore Dr. Chicago, IL
• Northwestern Campus
• Central downtown location
• Near carrier services serving Chicago loop
• Telephone switch room
• Colo space available
• Multiple carriers access
• Ameritech, AT&T, Qwest, Global Crossing, Global Crossing, MCI
Worldcom
• Other builds possible