[p2p-research] P2P Email
marc fawzi
marc.fawzi at gmail.com
Sat May 9 10:51:04 CEST 2009
Marco wrote:
"a) running the system on one's home or mobile computers, rather than
buying/renting for this very purpose a virtual server in a reliable
datacenter is such a bad idea for so many technical and other
reasons from social to environmental, reasons which DON'T change at
all whatever version of IP is used, that I don't think I'll ever do
it myself, nor I recommend it to anybody who uses email for more
than casual chat.
"
You're still missing the tunneling part and you brought up the static
IP part which is not part of the P2P Email architecture (that would be
dumb since it ties a client to a specific network access point)
IPv6 ALLOWS point to point communication with ease where it is very
difficult today. That is the whole point. Suggesting that we'd use the
static IP feature of IPv6 is very wrong in the context of a P2P Email
architecture.
Marco wrote:
"
>> In the best case, it means that you MUST carry that object along
>> with you every moment of your life, you can't ask a friend "let me
>> check my email, please" A nightmare.
>
> theoretically yes, one could introduce new protocols to add
> redundancy, security and robustness to such a scenario, or,
> individually, buy server-class hardware, UPS for backup power,
> subscribe to two or more ISPs for redundant connectivity, that is
> spend 10x or more each month than a virtual server costs. But why?
"
Completely wrong in the context of the commonly understood P2P architecture.
See preceding comments on NOT using static IP.
Wherever your connect from as long as you have an ID on the P2P
network you can be routed it to.
I'm suggesting that P2P Email is local to those using a particular P2P
client but if client developers agree on a standard then it becomes
global p2p email, since all clients will share the same distributed
user ID lookup inrastructure (think DNS like mechanism for user IDs so
instead of email being sent to a domain it is sent to a user ID) and
same transport protocol.
So I think the problem in this communication is the intuition of a
software engineer (myself) which is normally matched to the intuition
of other software engineers when discussing software architecture but
cannot be readily matched to the intuition of someone who is not
versed in software architecture and commonly understood patterns both
at the network and software architecture levels.
That is why it was misunderstood by Marco that I'm suggesting static
IP addresses be used for routing rather than p2p user IDs which would
be picked up right away by any person whose background is in network
and software architecture.
So with p2p email as I'm suggesting you are completely empowered and
redundancy is easily achieved in an IPv6 based network and
established architectural patterns exist and are used in p2p
architectures in all of the p2p file-sharing apps that exist today.
Also, email can be picked up from any PC or device running the p2p
client. It's just like email today except it is based on a p2p network
rather than centralized or distributed/federated network. The whole
domain name bullshit needs to go and people need to start accessing
web pages and email etc by P2P ID not by domain names. You ID is
everything you need and everything anyone needs to reach you in a true
p2p network.
Key issue for p2p architectures in Ipv6 environment is going to be
security (it already is in IPv4 but as p2p becomes more widely used
the security [of each client machine] becomes a bigger issue)
Marc
On Fri, May 8, 2009 at 11:10 PM, M. Fioretti <mfioretti at nexaima.net> wrote:
> (properly trimmed, quoted with standard markers, reformatted in
> natural reading order and with attribution restored, out of respect of
> future archive readers and other subscribers who may have jumped in
> right now, or be confused by the technical details)
>
> On Fri, May 08, 2009 20:00:34 PM -0700, marc fawzi wrote:
>> Marco Fioretti wrote:
>> > I take this as a confirmation that when you say "p2p mail" you
>> > mean:
>> >
>> >"everybody managing all their incoming and transmitted email
>> >directly and exclusively from THEIR OWN PERSONAL COMPUTER, ie the
>> >personal laptops, PDAs, whatever... that they always carry with
>> >them, or the personal desktop computers they have under their desk
>> >at home".
>>
>> Correct, in the context of my definition of "p2p" which is somewhat
>> purist when it comes to the p2p architecture.
>>
>> However, in that definition I include distribution for redundancy
>> and some centralized coordination (that can be distributed too for
>> redundancy) so, with respect to the latter, it's not completely
>> decentralized.
>
> OK, so this is where the difference and the source of former confusion
> is. Let's then restart and sum up a few other things for the benefits of
> our readers.
>
> To have a permanent email address, send and receive email with it, a
> generic Joe User needs:
>
> 1) an email address with a name (left of the @ sign) and a domain part
> (right of the @ sign): joe.user at somedomain.com
>
> 2) one or more computers (physical or VIRTUAL), directly connected
> 24x7 to the Internet, with FIXED numeric IP address(es), because IP
> packets are routed to IP addresses, NOT domain names. If the IP
> addresses weren't constant, your email correspondents would never
> know what your address will be next week, ie nothing would work.
>
> 3) some software, running on such computer(s) which sends, stores,
> receive email, filters spam, blocks viruses, sends autoresponders
> if needed, etc...
>
> NOTE: 2) and 3) together are, or should include, what is called a Mail
> eXchange (MX), that is Joe User "email management and routing system".
>
> 4) an entry into the worldwide Domain Name System telling everybody
> worldwide that the MX for somedomain.com runs at the constant IP
> address(es) mentioned in 2). Without this mapping, NOTHING works,
> because IP packets are routed to IP addresses, NOT domain names.
>
> this is the way it works today, and I'm focusing on this. Futuristic
> visions are good, but something else and a whole different thread, at
> least for me here and now.
>
> Now, let's look again at points 1-4 above. What happens today is that
> points 2-4 are centrally managed by corporations, who own the domain
> part of the address (gmail, yahoo, everybody else) and provide the
> service for money, advertising, etc... or as a business tool in the
> case of work email addresses.
>
> Here we all agree that this is bad: privacy concerns, lack of
> robustness, lack of flexibility, non standard configurations, no
> rights to change the terms of services, etc...
>
> Starting from this common ground, Marc Fawzi says IPv6 will make p2p
> email as defined above possible, because it will make available enough
> static IP addresses that every individual will be able to do points
> 2-4 directly without technical complications impossible to avoid today.
>
> In this vision, everybody will be able to associate to his static IPv6
> address to his own personal computers and run on those personal
> computers a new generation of email software integrating all I listed
> in 3) and then some, without intermediaries.
>
> All the comments from Marc about NAT, bypassing firewalls etc... that
> should disappear with the switch from IPv4 to IPv6 come from the fact
> that:
>
> *) he suggests or envisions that people could and should use their
> everyday (mobile) computers for point 2
>
> *) but (and he's right on this) with IPv4 (today) these computers are
> routinely assigned "bogus" IP addresses which are very hard, or
> impossible, to make them "look" as static, ie suitable for email
> routing. In other words, with IPv4 it's impossible to run your
> email from your own personal computer. I agree on this, of course.
>
> Now, I agree that everybody taking full control of their own email
> management is definitely a good thing. What I say is simply:
>
> a) running the system on one's home or mobile computers, rather than
> buying/renting for this very purpose a virtual server in a reliable
> datacenter is such a bad idea for so many technical and other
> reasons from social to environmental, reasons which DON'T change at
> all whatever version of IP is used, that I don't think I'll ever do
> it myself, nor I recommend it to anybody who uses email for more
> than casual chat.
>
> b) developing new software which would overcome all the problems
> created by using a cheap laptop as MX, and be simple to use is a
> very big task
>
> c) points 2-4 can be already done by individuals with IPv4 and
> existing software today. They're exactly what I, an individual, am
> already doing. What is missing to make worldwide p2p email is
> "only":
>
> - laws that don't make this illegal or affordable only by big
> corporations, maybe an ICANN reform
> - ISPs selling "individual MX virtual servers" with a GUI which
> spares non geeks from all the manual configuration work I had
> to do by hand. No need for more server or client software,
> what exists is adequate, even if it should always be improved
> - enough IP addresses for everybody to have one to associate to
> their "individual MX virtual server" package.
>
> but last point is exactly the first thing I said yesterday: the only
> reason I see why IPv6 should ever enter this picture is to have enough
> numeric fixed addresses. Because every other problem caused by IPv4,
> that is getting rid of NAT, working through ISPs or corporate
> firewalls,etc... is true but irrelevant compared to the cons of
> running a server from a personal computer. As I wrote yesterday:
>
>> Think Michel's laptop. Having a personal (= the one you carry with
>> you, keep at home etc... ) cheap computer being the single point of
>> routing and failure of your communications means that if it's
>> stolen, drenched with coffee, blasted by electrical charges or
>> breaks in any other way you're really toast. It means that if you go
>> for a few days without broadband email to you gets bounced, because
>> there's no other computer able to store it.
>>
>> In the best case, it means that you MUST carry that object along
>> with you every moment of your life, you can't ask a friend "let me
>> check my email, please" A nightmare.
>
> theoretically yes, one could introduce new protocols to add
> redundancy, security and robustness to such a scenario, or,
> individually, buy server-class hardware, UPS for backup power,
> subscribe to two or more ISPs for redundant connectivity, that is
> spend 10x or more each month than a virtual server costs. But why?
>
> Marco Fioretti
> --
> Your own civil rights and the quality of your life heavily depend on how
> software is used *around* you: http://digifreedom.net/node/84
>
> _______________________________________________
> p2presearch mailing list
> p2presearch at listcultures.org
> http://listcultures.org/mailman/listinfo/p2presearch_listcultures.org
>
--
Marc Fawzi
Facebook: http://www.facebook.com/people/Marc-Fawzi/605919256
LinkedIn: http://www.linkedin.com/in/marcfawzi
More information about the p2presearch
mailing list