Ben,

My feedback is embedded below.

Thanks,

--

JG

[cid87442*[email protected]]

James Gould
Fellow Engineer
[email protected]<applewebdata://13890C55-AAE8-4BF3-A6CE-B4BA42740803/[email protected]>

703-948-3271
12061 Bluemont Way
Reston, VA 20190

Verisign.com<http://verisigninc.com/>

From: Ben Schwartz <[email protected]>
Date: Wednesday, February 4, 2026 at 4:17 PM
To: "Gould, James" <[email protected]>, "[email protected]" 
<[email protected]>, "[email protected]" <[email protected]>
Cc: "[email protected]" <[email protected]>, 
"[email protected]" 
<[email protected]>, "[email protected]" <[email protected]>
Subject: [EXTERNAL] Re: Re: [regext] Re: draft-ietf-regext-epp-https-02 early 
Httpdir review
Resent-From: <[email protected]>
Resent-To: <[email protected]>, <[email protected]>, <[email protected]>, 
<[email protected]>, <[email protected]>, <[email protected]>, 
<[email protected]>, James Gould <[email protected]>, 
<[email protected]>, <[email protected]>, 
<[email protected]>, <[email protected]>
Resent-Date: Wednesday, February 4, 2026 at 4:17 PM


Caution: This email originated from outside the organization. Do not click 
links or open attachments unless you recognize the sender and know the content 
is safe.


________________________________
From: Gould, James <[email protected]>


>  EoH uses HTTP GET to the HTTP server to establish the EPP session that 
> returns the EPP greeting packet.  EoH then goes into a command and response 
> loop and leverages HTTP POST from the client to send the EPP command packet 
> and to receive the EPP response packet.



As I mentioned, this design seems unlikely to work well in common "cloud" CDN 
models, because the requests that make up one "loop" will likely be 
load-balanced across several EPP session servers.



JG-The HTTP GET establishes the HTTP session, that can be used to route to the 
appropriate EPP server for the command and response EPP flow.  EPP is a 
stateful application protocol and the EPP session is established upfront to 
help load balance to the correct EPP server.



> The goal with EoH is to enable the 20+ year old EPP application protocol to 
> ride on HTTP to provide for a more Cloud-friendly option for registries.



It would help to have a clearer sense of what "cloud-friendly" means here.  In 
my view, the proposed design is not "cloud-friendly", because it is likely to 
be broken if deployed in the most natural way using an HTTP gateway.



JG-Cloud-friendly means that Cloud HTTP gateways can be leveraged instead of 
requiring the registry to implement custom EPP gateways that is currently 
required for EoT.



> HTTP has been used as an EPP transport by many EPP registries, and we want to 
> standardize it in draft-ietf-regext-epp-https.



Does this draft document an existing deployed protocol, or is it proposing a 
new one?  If the protocol is already deployed and not subject to significant 
revision, then the proposed status should be "Informational".



JG-No, it’s based on a mix of past deployments and has been re-considered 
considering HTTP/2 and HTTP/3.



If the IETF has change control, then I suspect EPP-over-WebSocket would be an 
easier way to pass EPP through HTTP cloud infrastructure.



JG-That may be the case, but we still believe that EoH can be deployed 
effectively.





--Ben Schwartz
_______________________________________________
regext mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to