/
MMA Agenda and Minutes 11Aug 2020

MMA Agenda and Minutes 11Aug 2020

Owned by Bob Monkman (Deactivated)
Last updated: Aug 11, 2020
3 min read

Attendees

@Bob Monkman (Deactivated) (Intel)

@Scot Steele (AT&T)

@Beth Cohen (Verizon)

@Nick Chase (Mirantis)

@Beth Cohen (Verizon)

@Andrew McLachlan (Deactivated) (Red Hat)







Agenda Bashing



Agenda

Discuss Proposed Webinar Panel Session for 22 September for Intel Network Builders series with talking points and participants – 

Topic: Incorporating Hardware based Network Interfaces into Open Source Infrastructure Reference Models

Outline: Panel discussion with representatives from across the Telecom Industry ecosystem

  • Nick Chase – Mirantis Moderator

  • Beth Cohen – Verizon

  • Scot Steele – AT&T

  • Bob Monkman – Intel

  • Phil Robb – Ericsson



Start with quick 5-10 minute overview of CNTT (or whatever it will be called by then) and its mission -- Nick

Questions:

  • How does a SDN network reference infratructure model/architecture incorporate real hardware elements?

  • How can this be done without sacrificing vendor neutrality?

  • How does this affect the vendor/operator technical relationships?

  • How can the CNTT framework help?

Updated discussion on captured summary accomplishment talking points/Call to action in their areas that we can amplify in marketing messages – 

Review any update on Field trials.  – Mirantis is building the RI1 implementation based on the specifications from RA1

ROI Discussion - Brainstorming Nick & Scot- Update

Open Discussion on continued Marketing Messaging



      

Outstanding ARs

@Bob Monkman (Deactivated) begin drafting questions for participant survey

@Bob Monkman (Deactivated) ask Phil Robb for rep on panel



Minutes:

1st pass summation of accomplishments/next steps polling

Accomplishments:

defined an architecture w/ clear APIs, requirements and test requirements

defined profiles for compute-intensive , network-intensive workloads , capabilities and measurements and mapped them together

Common APIs, common language about expectations and descriptions of needed solutions, makes it easier to deliver on the expected solutions

common ways to express resources and easier to track changes,; easier to code from common specification language



We have built a generic, flexible and yet concrete model that can spawn multiple Ref architetures 

we have been identifying and closing gaps previously unaddressed 



The Ref Model and profiles and relationships are already making it easier to write RFxs by creating standard ways of asking for and assessing solutions

For some operators, the RA is useful for operations to build a concrete architecture while RI/RC is becoming a recognized sandbox for evolving the solutions



Meanwhile, we have engaged experts from a broad cross section of the supply chain and established strong working relationships with other stakeholders on OSS projects and SDOs

Note: TIP Open COre project engaged to consider there k8s implementation to be based on RA-2



We are now evolving the models for practical real world hybrid deployments of VM/container architectures to realize the necessary co-existence of IaaS/Caas which much be managed on the same HW w/ complex multi-tenant ecosystem

CHallenges/where we need help

  • how to integrate SDN into cloud arch

  • how to automate build, deploy and operational aspects

  • how to exploit HW capabilities in a vendor neutral way- Need Open HW Infra Mgmt



Revisited the discussion on the nomenclature of the main workstreams Rx-1 and Rx-2 ,   OpenStack/VM v Containers/k8s,  but since nothing is cut and dry , distinct

The real problem is that numeric distinction implies versions

RA- MS  is one idea where MS= Microservices and the RA-1 could be expressed as RA-VM  



Not much progress on the Costing Model due to focus on FMO 

Field trials  with AT&T and Mirantis still working thru preparation issues. 



Next steps: Continue to flesh out questions for Baraque release statements survey and milestones needed to prepare for the Spet webinar panel.