David,
We really appreciate your review and comments. Please see below for the 
resolutions.
Sorry for the delayed response. I missed yours when I was going through the 
comments from other reviewers.

The revision -23  
https://datatracker.ietf.org/doc/draft-ietf-rtgwg-net2cloud-problem-statement/ 
has addressed the comments from OpsDIR, RTGDIR, DNSDIR and GENART. Changes to 
your comments will be reflected in the -24 revision.

Linda
-----Original Message-----
From: David Black via Datatracker <[email protected]>
Sent: Monday, April 3, 2023 4:13 PM
To: [email protected]
Cc: [email protected]; [email protected]
Subject: Tsvart early review of draft-ietf-rtgwg-net2cloud-problem-statement-22

Reviewer: David Black
Review result: Not Ready

Transport Area Review:

        Dynamic Networks to Hybrid Cloud DCs: Problem Statement and
                           Mitigation Practices
              draft-ietf-rtgwg-net2cloud-problem-statement-22

Reviewer: David L. Black ([email protected]<mailto:[email protected]>)
Date: April 3, 2023
Result: Not Ready

>From a Transport Area perspective, there's not a lot of relevant content in 
>this draft.
Section 5 mentions IPsec tunnels, which raise the usual transport-related 
concerns in dealing with tunnels.  Those concerns can be primarily addressed by 
citing appropriate references, e.g., MTU concerns are discussed in the tunnels 
draft in the intarea WG, and ECN propagation is covered by RFC 6040 plus the 
related update draft for shim headers in the TSVWG working group.  I don't see 
any serious problems here.
[Linda] For the MTU introduced by IPsec tunnels, how about adding the following 
sentences?
      As described in [Int-tunnels], IPsec tunnels can introduce MTU problems. 
This document assumes that endpoints manage the appropriate MTU sizes, 
therefore, not requiring VPN PEs to perform the fragmentation when 
encapsulating user payloads in the IPsec packets

IPsec tunnels are over public internet, which doesn’t support ECN. Why need to 
mention RFC6040?


OTOH, from a broader perspective, the draft is not a coherent problem statement 
- it discusses a plethora of technologies ranging from MPLS to DNS, often 
without making any connections among them (e.g., section 6 identifies policy 
management as a requirement, but there's no discussion of policies that require 
management elsewhere in the draft).
[Linda] This document describes the network-related problems enterprises face 
when interconnecting their branch offices with dynamic workloads in third-party 
data centers (a.k.a. Cloud DCs) and some mitigation practices. It is a list of 
technologies ranging from VPN to DNS.


I'm not even sure what the scope of the draft is, e.g.:

a) The abstract states that the draft is "mainly for enterprises that already 
have traditional MPLS services and are interested in leveraging those 
networks," but section
3.4 discusses 5G Edge Clouds, which are rather unlikely to use MPLS.
[Linda] The document is mainly for enterprises that already have traditional 
VPN services and are interested in leveraging those networks (instead of 
altogether abandoning them). MPLS (which is now replaced by VPN) is just one 
example.


b) There are at least three roles for BGP in this draft that are not 
disambiguated - IGP, EGP, and VPN routing protocol for MPLS-based VPNs, e.g., 
EVPN.  Section 4 would be a good place to clarify this by describing the 
Gateway interfaces in detail, including the role of BGP.
[Linda] Connecting to Cloud needs BGP, but doesn’t run IGP, EVPN.
The intend of the draft is to identify future work in BGP.

In its current form, I don't understand the target audience or purpose of this 
draft, especially the head-spinning mixture of topics in section 3, so I cannot 
recommend IETF publication of the draft in its current form.
[Linda] The intent of the document is to lay out current mitigation methods and 
additional work on extension to BGPs, such as 
https://datatracker.ietf.org/doc/draft-ietf-idr-sdwan-edge-discovery/

Perhaps the draft ought to be focused and organized around extending and/or 
using MPLS and MPLS-based VPNs - much of the material in Sections 4 and 5 would 
be applicable, and some of the worst of section 3's distractions (e.g., 5G, 
DNS) could be avoided or at least scoped to the relevant VPN technologies.
[Linda] DNS issues introduced by connecting to Cloud DCs were strongly 
requested by DNSOps and OpsDIRs.

Thank you very much
Linda



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

Reply via email to