How important is reverse DNS for websites?

[jay5r]

I understand why reverse DNS (PTR records) is important for IPs used by web clients and when sending email, but if an IP is only used for web serving is there a reason why the PTR record would be important?

In the world of IPv4 there were (are) a bunch of websites on one IP, so it kinda made sense - do it once and whatever benefit it has applies to a bunch of websites. But in the world of IPv6 where every website can have its own IP, it's just laborious and if the benefit is non-existent, then why bother?

[phranque]

this subject is extensively discussed in this IETF document:
Reverse DNS in IPv6 for Internet Service Providers [tools.ietf.org]

[jay5r]

phranque - Thanks, but that document argues sides of the issue to the point of being self-contradictory. It mostly covers how it might be done while acknowledging that most of what it suggests isn't currently possible.

It repeatedly mentions that forward and reverse names should match - but that doesn't happen now in web serving environments where one IP serves many hosts since you can't have multiple reverse DNS responses. And if the forward DNS is a naked domain - if my memory serves me correctly, reverse DNS isn't supposed to return naked domain names.

But the IETF document does mention that "the need for a PTR record and for it to match is debatable as a best practice". And none of the cases mentioned in the article where a reverse DNS is useful/needed apply to an IP that's only used for web serving.

In the end this comes down to the capabilities of my host. Two years ago they couldn't even do a PTR record for IPv6 (though they did solve that problem a month after I requested it). Based on that IETF document I'm not sure how much I'll push the issue (especially right at the moment since most of their first line tech support guys are in Ukraine and have bigger problems to deal with). In the end it sounds like a "nice to have", but not a "must have".

[phranque]

Thanks, but that document argues sides of the issue to the point of being self-contradictory. It mostly covers how it might be done while acknowledging that most of what it suggests isn't currently possible.

exactly why i referred you to the document instead of giving a black-and-white answer.
keep in mind that the state of IPv6 implementations may have advanced in the 4+ years since this was published...