Zongning
2014-08-14 02:41:57 UTC
Hi, all,
Please see below minutes of Toronto BoF. Thanks to Jon Hudson for taking notes. :)
==================================================
Goal of Today is to achieve Rough Consensus on VNFPool Charter
Sue Hares - VNFPool Use Cases
Based on a personal experience deploying VNFPools
Question -> No
Problem Statement - Melinda Shore Speaking
Main Changes since 89 reviewed. Main focus has been on clarification.
There seems to be a significant concern to the VNF
Challenges and Open Issues
Redundancy Management
Interaction Between a VNF and a Service Control Entity
Reliable Transport
Questions->
(TBD_Human_01) Is there accounting for stateful flows?
Melinda: Currently we are not, and it may be handled in a different layer. But please if anyone has ideas on how to solve any of the stateful issues please submit them. This is a very challenging problem.
Generic Use Case (Masaki Fukushima)
(TBD_Human_02 Will Physical and Virtual devices be considered?
Masaki: No we are currently only looking at virtual devices.
F5 Guy: I think this is a great goal, but I don't see how you can provide any real redundancy without stateful capabilities?
Melinda: For now this is out of scope, but does not need to be in the future.
vEPC Use Case (Marco Liebsch)
Questions-> No
vCDN Use Case (Oscar Gonzalez De Dios)
TBD_Human_03: (missed the question)
Yes we have the load balancing today using different cacheing and different ways of moving and providing resources.
TBD_Human_03: Do you see this being a CND solution.
Oscar: It's not application specific, I think it's more general. Daniel do you have an opinion?
Daniel King: The load balances has been a contentious issue. We really need to nail this down, perhaps even address it in the Charter. But there are many options being discussed. I will gather the options and send them all on the mail list for discussion.
On to the Charter:
(New Guy) Is there any redundancy planed for the manager?
(New Guy) I would like to add my view that
Nichan, Huawei - Why don't we use TCP/IP? Why are you trying to invent a new transport protocol
Melinda:
TBD_Human_04: At the end of this working group are we expecting one single protocol produced? Or would this be more a framework allowing multiple solutions.
David Allen: Do you plan to alining this architecture with the ETSI NVF
David Allen: What will you do if your gab analysis results in no gaps?
Melinda: We don't expect to have any lack of gaps.
David Allen: Well I think you may be surprised once you look at what is available in the open source market
Ericcson_Guy: Are you trying to standardized a hypervisor or trying o change anything on the host side.
AD: We do protocols, we don't influence hypervisors or host operating systems.
Ericcson_Guy: I just don't see the value being added, when you look at what is in the market
Kevin Riverbed: Is there any consideration for different performance characteristics and how selection is done and if so can you document it?
Melinda: Yes, we don't have much on that yet. As always we need text. So please volunteer what you have.
TBD_Human_05 - Missed all of it.
Oracle_Guy:
Melinda: Point taken I think that is essentially correct.
TBD_Human_06: Is discovery of VNFs or VNFpools being included in the charter?
Melinda: Yes, that is something being looked at elsewhere and we'll need to look at it and see what we can actually deploy.
TBD_Human_06: Missed the comment.
Daniel King : Just following up on Dave's comments about ETSI NFV. Have we look at a formal liaison to help make that happen?
Christopher: I have a concern that just because we are grouping existing objects in a new way that a new protocol is needed. Pool Capabilities vary quite a lot between Containers and Orchestrators and it's all over the place and coordinating that with our work is a big worry.
Francois Cisco: I want to restate my concern about interoperability and also my interest in finding a generic solution, or something more specific. Both methods have good and bad but have serious impact.
======================================================
Please let us know anything you would like to correct, or comment. Co-chairs will work out some action items (under the supervision of responsible AD) as BoF follow-up shortly.
Thanks.
-Melinda & Ning
Please see below minutes of Toronto BoF. Thanks to Jon Hudson for taking notes. :)
==================================================
Goal of Today is to achieve Rough Consensus on VNFPool Charter
Sue Hares - VNFPool Use Cases
Based on a personal experience deploying VNFPools
Question -> No
Problem Statement - Melinda Shore Speaking
Main Changes since 89 reviewed. Main focus has been on clarification.
There seems to be a significant concern to the VNF
Challenges and Open Issues
Redundancy Management
Interaction Between a VNF and a Service Control Entity
Reliable Transport
Questions->
(TBD_Human_01) Is there accounting for stateful flows?
Melinda: Currently we are not, and it may be handled in a different layer. But please if anyone has ideas on how to solve any of the stateful issues please submit them. This is a very challenging problem.
Generic Use Case (Masaki Fukushima)
(TBD_Human_02 Will Physical and Virtual devices be considered?
Masaki: No we are currently only looking at virtual devices.
F5 Guy: I think this is a great goal, but I don't see how you can provide any real redundancy without stateful capabilities?
Melinda: For now this is out of scope, but does not need to be in the future.
vEPC Use Case (Marco Liebsch)
Questions-> No
vCDN Use Case (Oscar Gonzalez De Dios)
TBD_Human_03: (missed the question)
Yes we have the load balancing today using different cacheing and different ways of moving and providing resources.
TBD_Human_03: Do you see this being a CND solution.
Oscar: It's not application specific, I think it's more general. Daniel do you have an opinion?
Daniel King: The load balances has been a contentious issue. We really need to nail this down, perhaps even address it in the Charter. But there are many options being discussed. I will gather the options and send them all on the mail list for discussion.
On to the Charter:
(New Guy) Is there any redundancy planed for the manager?
(New Guy) I would like to add my view that
Nichan, Huawei - Why don't we use TCP/IP? Why are you trying to invent a new transport protocol
Melinda:
TBD_Human_04: At the end of this working group are we expecting one single protocol produced? Or would this be more a framework allowing multiple solutions.
David Allen: Do you plan to alining this architecture with the ETSI NVF
David Allen: What will you do if your gab analysis results in no gaps?
Melinda: We don't expect to have any lack of gaps.
David Allen: Well I think you may be surprised once you look at what is available in the open source market
Ericcson_Guy: Are you trying to standardized a hypervisor or trying o change anything on the host side.
AD: We do protocols, we don't influence hypervisors or host operating systems.
Ericcson_Guy: I just don't see the value being added, when you look at what is in the market
Kevin Riverbed: Is there any consideration for different performance characteristics and how selection is done and if so can you document it?
Melinda: Yes, we don't have much on that yet. As always we need text. So please volunteer what you have.
TBD_Human_05 - Missed all of it.
Oracle_Guy:
Melinda: Point taken I think that is essentially correct.
TBD_Human_06: Is discovery of VNFs or VNFpools being included in the charter?
Melinda: Yes, that is something being looked at elsewhere and we'll need to look at it and see what we can actually deploy.
TBD_Human_06: Missed the comment.
Daniel King : Just following up on Dave's comments about ETSI NFV. Have we look at a formal liaison to help make that happen?
Christopher: I have a concern that just because we are grouping existing objects in a new way that a new protocol is needed. Pool Capabilities vary quite a lot between Containers and Orchestrators and it's all over the place and coordinating that with our work is a big worry.
Francois Cisco: I want to restate my concern about interoperability and also my interest in finding a generic solution, or something more specific. Both methods have good and bad but have serious impact.
======================================================
Please let us know anything you would like to correct, or comment. Co-chairs will work out some action items (under the supervision of responsible AD) as BoF follow-up shortly.
Thanks.
-Melinda & Ning