I was thinking about the discussion in the Ops Area today about whether DNS Ops 
would be more productive if split into ops and ext.

It seems to me that the problem with prioritizing the work in the DNS area is 
due to a lack of a shared understanding of the requirements and some 
fundamental disagreement about their relative priority. There are operators who 
are not convinced of the value of deploying some of the stuff that has been 
developed (e.g. DNSSEC) and there are developers who are not convinced of the 
wisdom of some of the things operators want (e.g. EDNS-client-subnet). 
Generally the operators want to reduce the risk of service impacts, and 
increase service differentiation so they can find some way to monetize their 
services. The developers are looking for simplicity, elegance and robustness in 
the software and the protocol, and to avoid future CVEs. Sometimes these 
interests overlap, but the relative priorities are rarely in synch. 

When a new draft is proposed, there is not usually much discussion of the 
business problem being addressed, and when you look at the technical problems 
being addressed, often they benefit one party or another in the DNS unevenly.  
Generally, it isn’t explicitly discussed who the work is ‘for', or how 
widespread or impactful the ‘problem’ being addressed is. Some documents 
languish because some people view them as a waste of time. If it was more 
explicit about who each draft is intended to benefit, it might be easier to 
‘take turns’ at the well. 

More attempts to estimate the scale and impact of the problem being addressed, 
along with any data that supports that estimate, would help. It does seem 
sometimes as if there are competing workstreams (enabling DNSSEC, enabling 
encryption, enabling IPv6 transport, enabling new applications) and if those 
were more explicit then maybe the WG(s) could attempt to balance them. 

I don’t seen any long-lasting solution to this. It would be easier to 
prioritize if the work were divided into operational issues and protocol 
extensions, but eventually the ops wg could end up with a lot of drafts without 
implementations, and the ext wg might end up with a lot of protocol extensions 
that aren’t deployed. 

_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to