NETF-1991-03-21 KTH-Stockholm High speed services / Real time video pilots HSN RTVP NETF-HSN@nic.nordu.net NETF-RTM@nic.nordu.net A combined workgroup on topics of high speed networking, future NORDUnet network structure, what gigabit eating services could be used, etc. Attending also Mr. Kevin Moran of Sprint International who earlier same day gave a presentation of their Frame Relay services. Participants: H}vards Eidnes Havard.Eidnes@runit.sintef.no Kevin Moran (SPRINT) UN:K.Moran;O:Sprintintl;A:TMailUK;C:GB Harri Salminen hks@funet.fi Matti Aarnio mea@funet.fi Juha Hein{nen jh@funet.fi Petri Ojala Petri.Ojala@funet.fi Mika Uusitalo uusitalo@tut.fi Geir Pedersen Geir.Pedersen@use.uio.no Reijo Metso gucre@gd.chalmers.se (guere ?) Gunnar Lindberg lindberg@cs.chalmers.se Gardar Nielsen goggi@rhi.hi.is Rolf Nordhagen Rolf.Nordhagen@use.uio.no Bernhard Stockman boss@sunet.se Mats Brunell matsb@sics.se Bjarne ???? ?? @ ?? Hans Eriksson hans@sics.se and others who were listening discussion or didn't write their names to a circulated paper. Chair: Hans Eriksson Secretary: Matti Aarnio Discussion started by Juha Hein{nen's views of network products, what we do now, what COULD be achieved in under a year if PTTs get in to the action, and what is "talking about the moon". Juha: We have now a stockholm centered star which is susceptible to single faults. (mea: Local loop from STI to KTH is SINGLE line using TDM for various subchannels. Have a fault on that, and whole NORDUnet breaks apart immediately. ) Juha: We must get a physically fully redundant mesh network. (Preferrably creating full connectivity among Stockholm, Helsinki, K|penhagen, and Oslo. ) Using US Sprint Frame Relay connection CLOUD to connect USA (at least two points) and NORDUnet (with at least two connections to cloud) plus whoever of central europe wants to join in. Juha: By September-91 there should be FR service available (not from US-Sprint in nordic countries, but from other - PTT? - sources. [mea]) We need attractive pilot proposals to be given to US-Sprint and PTTs. Bjarne: EEPG believes that isosynchronous service is necessary (for video confrencing etc -- as long as bandwidth is large enough, transport ends can do buffering to get local dataflow to be constant, but network need not to be such.) (Also, isosync is still much of a question of future developements.) Mats: Pilot (network) must be able to support at least our current services. Juha: We need to be pilot customers to PTTs and other vendors so they will start to implement FRs etc. (Juha?) Also we are able to identify ourself as only large nordic organization whose networking solutions and technologies are followed closely, and who can promote new technologies (like FR) to smaller [and possibly more conservative] customers (like businesses.) (Moran of Sprint smiling and nodding...) Chair: We should start deploying PhoneTalk etc service pilots to people. Rolf: Distant/distance education Chair: Co-authoring! Chair: Rolf's group (USUP) will get users to use all these gigabit swallowing services. Juha: HSN group will need to produce a "selling" document with buzzwords for politicians (like Distance Education.) Chair: On 12th April in Coepenhaken there will be a meeting with PTT persons on networking (no politicians). This group need to produce within 2 (two) weeks: - A technical draft of networking infrastructure -> Mats + Juha - Motivation ("needs paper") by USUP and VAS groups Usage & Users ( Rolf + Mea + Hks ) Inventory of: - existing HSN - planned HSN - needed HSN How to deploy any multimedia pilots ? What there is ? - PhoneTalk - PicturePhoneTalk - ISDN as carrier for something ? - ST2 as carrier for something ? To whom to deploy them ? Or who does something with MultiMedia ? PhoneTalk -- 1/2 hour work on Sparc. KTH-SICS, UiO, Finland ? Co-Authoring Lund has some multimedia going on using 2Mbps codecs (dentisis etc.) UiO has multimedia lab, but no network infrastructure (coming soon, maybe even 34Mbps?) Rolf: Look for Mac users for potential on video confrencing. Mea: Space/Astronomy/Physics need tremendous bandwidths for remote experiments. (Not everybody wants to go to top of the mountain for one weeks nights to make measurements -- technicians are for that ;-) ) Hks: We must make a reservation of capacity for network management, othervice these PicturePhoneTalks will kill our network in no time flat. (By not leaving any bandwidth for management, and saturating all lines.) Chair: Lets not jump straight on to gigabit speeds on NORDUnet, as step from 64kbps to 1Gbps is so huge. Better to do intermediate steps and to fill every of those. Carrier and switching technologies for each speed class are less or more different and demanding when speed increases. Mats: FR cloud initial speeds should be at least 2Mbps. Chair: PicturePhoneTalk should fit into 384kbps by using good state of the art codecs. (In live, fairly hight quality.)