Re: Response to HTTP2 expresions of interest

In message <CAAbTgTv4QxwyBy5Fp5xg7A_WAQ2BAxrK=Ui932amJrXZ2iA50A@mail.gmail.com>
, Brian Pane writes:

>>From the perspective of a load balancer, having just those three fields in
>cleartext isn't sufficient. Sending a request to the proper upstream
>destination may require information from Cookie, X-Forwarded-For, and more.

(X-)F-F makes sense.

Cookies: not so, whenever people use cookies, they are working around
lack of session concept in HTTP.  HTTP/2.0 should fix that, so cookies
go away.

>I'm not too concerned about load balancers having to decrypt messages,
>though: SSL termination has been a key selling point for load balancers for
>many years.

That's not the same as it being a good idea.

Hosting providers are often unable to deploy load-balancers and
DoS mitigation, exactly because it would require them to have all
their hosted clients certificates.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

Received on Friday, 13 July 2012 19:51:06 UTC