AVCI
July 22, 2010
9:00 AM
Participants: Ben Fineman, Terri Saarinen - scribe, Simon Horne (Australia), Walt Magnussen, Charles Ganshorn (video), Laurie Kirchmeier (video), Ludwig Ganter at VA Tech, Zenon Mousmoulas from GRNet in Athens, GREECE, Brett Sweeny
AVCI Next Steps:
At this time we have had a chance to discuss some options that may lead to a dial plan that has the ability to be more or less universal. While it may not be possible to ever have anything be able to connect to everything in the world, it would be beneficial if at least a limited dial plan was not a part of the problem.
In terms of devices that should be supported, the group has agreed that the list should include at least:
• Any device on the PSTN
• SIP Voice
• H.323 Video
• SIP Video
• Telepresence Video
It is understood that communicating between disparate systems will result in least common denominator communications and may require transcoding and/or signaling protocol gateways.
A global solution appears to require three architecture components, being:
• Dial Plan - this involves the defining digits/characters that would be dialed in order to complete a call. Options would include:
o URI dialing utilizing DNS
o GDS dialplan
o E.164 utilizing existing campus numbers
o E.164 with a combination of campus numbers and E.164 Global Service numbers assigned to Internet2 by the ITU
• Devices that would contain global route plans in order to simplify end user dial plans. Options include:
o Peer-to-Peer dialing with no route aggregation
o Gatekeeper peering
o SIP redirect server
o Cisco IME server
• MCU or bridge to support multiple connections and transcoding when needed. Options include:
o Use campus MCUs and register them by assigning and documenting an address from which ever dial plan is selected.
o Have Internet2 provide and support a central MCU for supporting interoperability.
From this list of components, one could define a matrix that would list the service connections between two dissimilar systems and specify which connections would be supported and which would not. It is my recommendation that the next steps include:
• Have a sub group define the matrix possibilities and from there specify supported combinations.
• Have the sub group go back to the larger group for approval
• Have the sub group do a deeper dig of the solutions that seem to support the most connectivity. For these options we would document:
o Architecture components including cost estimates.
o Issues that would have to be resolved.
o Pros and Cons of this solution.
• From this step, we would have the group come up with a recommended solution(s)
• Submit a written recommendation to Internet2 leadership.
If the group agrees that this is the path, I would like to schedule a track session for the Fall Members meeting in November. This would mean weekly meetings for the subgroup for the month of August.
Walt opened the discussion regarding thought process. Ben did a recap of Walt's summarization of list of requirements for a dial plan, devices supported, architectural components and suggested way to move forward by forming a subgroup to work to define matrix of how different solutions would work with our requirements and bring back to group for review and recommendations.
Ben started by talking about whether or not the components on the list are condusive of what the group wants to see.
Supported devices:
• Any device on the PSTN
• SIP Voice
• H.323 Video
• SIP Video
• Telepresence Video
?? Is it possible to define all of the SIP options together? Walt mentioned visiablity on the network. If university is completely SIP compliant, he thinks this would work. If a campus has an option 11 or Meridan PBX coming off gateway, the switching architecture wouldn't deal with SIP video. He talked about running a call manager to manage dial plan for campus. Charles talked about call managers, trunking, and feeding in to SIP trunk - that only accepts audio codex. Accepting video is often not implemented.
Charles talked about the system vs. the SIP offerings, and how the interconnect between providers gets dumbed down to voice only.
Walt explained why SIP voice was separated from SIP audio. ?? said it's not that it can't interoperated with SIP, it's the SP offering doesn't offer that service - or provisioning isn't choosen.
?? talked about 10 digit dialing on his campus. If numbers assigned to him go to voice switch, least common denominator is what the switch can manage. More to do with how current numbers are routed and where they are assigned.
Maybe SIP is the signaling technology and the dial plan needs to provide for SIP. Walt talked about getting global numbers assigned to the group. If campus has SIP proxy assigned to handle any media stream or call, then there is no reason to assign global numbers.
Charles talked about Tanberg GCS mapping to E164, dialing name at domain, and alias that to gate or call agent to get to destination end point. Two wrinkles - native SIP endpoints only let you dial digits, not @domain. He said his UIs can't make that call.
?? talked about mapping E146 number to the UI. Charles said UI dialing takes determinig authority to mapping out of the picture. Some endpoints don't do the right look up unless they have alpha numeric string. Name@domain and then alias that in his own dialing plan.
Walt explained his reason why it's necessary do the matrix. Should it be done? Can it be done? Caveats of doing? What will it take? Walt reminded that we don't want to write the conclusion without doing the research. We all have our favorites, we need to logically map out the benefits and issues. What is the best way to get there? Do we want to present this at FMM10? If so, we need to put together a meeting for each of the four options, brainstorm workability and limitations.
Is it premature to put in placeholder for FMM10? Ben said timing is good, deadline for FMM10 is July 30? TO DO: Ben to submit FMM10 side meeting/track session proposal.
Walt asked the group if matrix can be done in August, with review in September, edit time in October and present it Nov. 1-4 at FMM10?
Walt asked who at Internet2 needs to be involved? Does this go to the council? AOAC? ICO? TO DO: Ben will investigate process.
Brett reminded callers that this is an official Internet2 Special Interest Group.
Walt recommended a sub group. In two weeks - go to a weekly call in August to talk the four options and bring the whole group together. Walt would appreciate for the first ones - URI - that folks we have from International areas be on the calls since they have more enum experience.
Zenon talked about a project, GEANT Educonf - no pointer for contact or person to suggest, perhaps, since you mentioned how other NRENS and other regionals are looking at these issues, a small project not yet started. He suggested someone from that group join a call. Walt said this would be great as we want to gather as much as we can.
Garr at Italian NREN may be able to get the group in-touch with someone. Zenon will search for more specific contact and send it to Walt. To DO: Zenon.
Walt will draft matrix for each of the four dialing plans we're looking at. This will lead to other questions - what architecture you have behind it, mapping devices, or have everyone put everything in all of their tables, or a central device to do transcoding, or do you count on the universities themselves to do the transcoding. Not sure if we have someone to manage large servers centrally.
Thoughts? Comments? Silence is lack of objection. Walt hopes, looking at matrices we see if universities should support or if we support centrally.
Let Ben know if you want to sign up for subgroup. Or if no one volunteers, we keep the group together. Ben thinks it would be great if folks step up. Prefers a volunteer army.
TO DO: Ben to schedule upcoming calls.
Once matrices are done, need to have a couple hour meeting to review and provide recommendation. Walt expects no clear winner and some contention.
Does this time work? August big vacation month - Walt suggested doing enum matrix 3rd or 4th week. Simon good with 11:00 PM.
Call end 9:45 AM