Chris,

It is very unfortunate that I had audio problem in Meetecho at today's RTGwg 
session. I can't hear from the Meetecho when I talk. I am not sure when my mic 
was muted.


After IETF106, we have made significant improvement. Removed all reference to 
SDWAN. And have the drafts mainly focusing on the problems associated with 
interconnecting branch offices with dynamic workloads in Cloud DCs, 
specifically  on problems that need additional work in IETF Routing area.

Based on the feedback from NANOG and DNS WG and DNSOps WG, we added the 
sections to describe the problems for DNS for workload in Cloud DCs:

>       DNS for Cloud Resources
*       Need to establish policies and rules on how/where to forward DNS 
queries to

      Cloud's DNS can be configured to forward queries to customer managed 
authoritative DNS servers hosted on-premises, and to respond to DNS queries 
forwarded by on-premises DNS servers.
*       Collisions can still occur. Better to use the global domain name even 
when an organization does not make all its namespace globally resolvable
>       DNS based solution to reach App Instances in the optimal Cloud DC 
> locations (Cloud discovery)
*       Dependent on client behavior
*       Client can cache results indefinitely
*       Client may not receive service even though there are servers available 
(before cache timeout) in another Cloud DC
*       No inherent leverage of proximity information present in the network 
(routing) layer, resulting in loss of performance
*       Client on the west coast can be mapped to DC on the east coast
*       Inflexible traffic control:
*       Local DNS resolver become the unit of traffic management


Main Network issues of Site <-> Cloud & Cloud <-> Cloud include:

>        Problems associated with Multiple Cloud DC Interconnection
         Different Cloud providers have different access method.  Today you 
have to hairpin the traffic to customer GWs
>       Multiple types of connections to workloads in a Cloud DCs
*       it is not visible to Apps in a Cloud  DC what type of network access is 
used.
>       IPsec P2P doesn't scale well with Multipoint mesh connection & poor 
> performance.
>       unknown segments --> difficult to collect end to end performance metrics
>       Problems of MPLS based VPN extending to Hybrid Cloud DC
*       PE might not have direct connections to Cloud DCs



Can you please provide some concrete feedback on drafts? Or request other Areas 
Directorate to provide feedback?

We greatly appreciate your help and guidance.

Sincerely,

Linda Dunbar

_______________________________________________
rtgwg mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/rtgwg

Reply via email to