Re: [rtcweb] Solutions sought for non-ICE RTC calls, not +1 (Re: Requiring ICE for RTC calls)

[Roman Shpount <[e-mail]>]

On Wed, Sep 28, 2011 at 3:10 PM, Olle E. Johansson <[e-mail]> wrote:

> If you look into your web browser, you will see that more and more sites
> are moving to only HTTPS. There's a reason for that. Facebook, Twitter and
> many others. Large sites that are not "banks", but social media or just
> information sites  like www.iis.se.
>
> When HTTP was written there wasn't CPU enough to handle mandatory SSL. Now
> we have CPU power to handle it and encryption accelerators are much cheaper.
> The world is different and comparing  the situation we have now with what
> was reality when RFC 2616 was written is a bad comparision that I would not
> consider valid.
>
>
This is not about CPU power -- the CPU impact of SRTP is a lot smaller then
the CPU impact of a lower bitrate codec. The problem with SRTP is that
debugging becomes much harder and vast majority of the current VoIP
monitoring tools will stop working. You can no longer use a simple packet
capture to create call logs and quality metrics for the voip call. You
cannot even do it for a particular client that is experiencing problem. AFAK
service provider can turn off encryption on GSM calls for troubleshooting
purposes. I think RTC should be able to do the same. Doing interop, if all
you have is SRTP, would become much harder, even if this is an interop
between two web browsers.
_____________
Roman Shpount