one-to-many messaging

[Ian Batten]

On 31 Mar 08, at 1445, Roland Perry wrote:
> In article <024e01c8932f$4a6e7cc0$e57ea8c0@Jinja>, James Firth <james2@jfirth.net 
> > writes
>> On Mon, 31 Mar 2008 13:29:00, Roland Perry wrote:
>>> I wonder why such an easy-sounding inspection of traffic data was
>>> regarded as "too difficult" by ISPs, before BT came up with its  
>>> rather
>>> complex platform to implement "Cleanfeed"?
>>
>> And now I don't quite see what you're saying.  From what I  
>> understand,
>> Cleanfeed was necessarily complicated for both performance reasons
>
> ie It's not as simple as just snooping the traffic data.

Hmm, I suspect it's also a Moore's Law and price of silicon issue.  At  
one point you could shove the entire outbound traffic of a consumer  
ISP through a cluster of Squid proxies, because more than one ISP did  
just that.  What squid does with each request is more complex, more  
resource intensive and more invasive than matching complete GET  
requests against a stop list, so at the time people were Squid-ing  
their entire peering load they could equally well have matched against  
an IWF list.

At the time Cleanfeed was first mooted, ISPs were _not_ Squid-ing  
their entire workload.  I don't know if this was for performance, cost- 
effectiveness, legal or other reasons.  It could be any of those.

But today, what Phorm are proposing is, again, at least as complex,  
resource intensive and invasive as Squid-ing the entire workload.  Of  
course, we don't know if Phorm's solution will actually work as  
advertised, because they haven't as yet had access to a complete  
network, but it's reasonable to assume it scales horizontally and  
therefore it's just a matter of throwing more boxes at the problem.

 From this I deduce that if Phorm is performant, a Cleanfeed solution  
which doesn't involve the complex two-stage architecture of only  
looking at the entire URL of things that match an IP-stoplist would  
today be possible from a technology perspective.  So the options are:

* When Cleanfeed was first done, the technology curves were different;  
or

* When Cleanfeed was first done there were regulatory, legal, IWF or  
other reasons that meant they couldn't just match all URLs against a  
stop list; or

* Someone had a neat idea to do it elegantly (Cleanfeed) and didn't  
need to look at the brute force way of doing it.

None of us know, or if we do, we're probably keeping quiet about it.

ian