[Top] [All Lists]

Re: Stranded postmasters

2009-12-05 03:59:52

Alessandro Vesely wrote:

Do RFCs contain more invitations to get out of the strand?

If you are referring to "stranded" as a result of complexity, this particular DKIM document is probably a good example of keeping them in the "strand."

I'm not sure whether it is the complexity of DKIM --or the mail system as a whole, for that matter-- that refrains postmasters from participating to discussions. Which discussions? Let me try and enumerate their types, AFAIK:

* IETF or IRTF related mailing lists,
* tool-specific *-user mailing lists,
* technique-specific *-discuss mailing list,
* DNSBL or rfc-compliance web sites,
* provider-specific feedback-loop web sites,
* generic web sites that also discuss mail issues.

It seems a lot of places. However, is the total number of subscribers comparable to, say, 25,665,515 (the "Total domains checked" on today)? I'd guess the percentage of "active" postmaster is around 1~4%, and the rest is stranded "inertial mass."

Alessandro, I am not entirely sure what the purpose of this discussion. But what is "Stranded?"

Your first comment seem to reflect complexity which my quip was regarding "free" vs "commercial." People paying for packages that attempt to make it "easier." There is no concrete correlation but commercial packages exist for a reason. Your second comment seem to reflect a sense of apathy, boredom or that "everything is working." The system is in autopilot, why break it? It may be a little of both.

But overall define postmaster? What is a "postmaster" today vs yesterday?

Today, I will venture that the disciplines are merged, especially with the help of integrated packages, systems and automation. I will tell ya from our experience, the pay scale of those administrating many of our installations are lower than they use to be. System Operator (sysop) retirement is one of the principle factors with our ~25 years old package. There are times where we have to "teach" SMTP, LIST SERVER or FTP because of the lower discipline levels. I use the old idea that our customers are not "unix wienies" and don't edit configuration files or don't want to - they want GUI (and today, they want it via the WEB interface whether that is Online or Offline).

There is also the factor that many people don't get involved until necessary, i.e. a problem perhaps. AVS was a big factor (no, the main factor) for me to get involved back 2003. Before that I put my trust and faith in the engineers involved with IETF protocols. I still do for the most part but no longer care to get involved in the the mental battles any more. My own IETF WG experiences has not been a charm. For that, I think another reason is that there operators (vs developers) seem to overwhelm the groups here. So perhaps thats another reason.

With SPF, I think you need to look at high value domains, i.e., companies, the fortune 1000, the banks, etc. Most have SPF domains.
SendMail has some good stats here:

According to them, 90% of the Fortune 1000 have SPF domains. 99% of the US banks have SPF domains. The levels for DKIM are pathetically low to nil. Now you have to ask why? A bad survey? Maybe.

But IMV, SPF was simpler to implement. You didn't need a project "TEAM" per se to get SPF started. DKIM is too complex at many levels and automation is probably the only thing that can help it. Does that mean "Stranded" sysops were somewhat interested but punted? The only way our customers will get to use DKIM is if we offered it to them and we held their hand thru the setup, configuration and automation of it all.


Hector Santos

<Prev in Thread] Current Thread [Next in Thread>