[Din] Re: next steps for DINRG Re: 2024110211733.AYC Re: 202410191733.AYC

[William Lehr <[e-mail]>]

Agreed that a centralized mechanism could be subject to distributed 
control.

Also agree that PSTN outages typically rarer than Internet outages, even 
locally, but having multiple networks with heterogeneous reliability 
performance can provide benefits. VoIP helped after Katrina.


On 10/23/24 9:55 AM, Abraham Y. Chen wrote:
> Hi, William:
>
> 0)    Thanks for injecting your thoughts. I am no linguist. So, I may 
> get mixed up with the meaning of certain words or expressions. 
> Hopefully, we can clarify each relevant ones and define them clearly 
> through discussions for the purpose of this Group.
>
> 1)    " ... A distributed mechanism may be subject to centralized 
> control. ...   ":
>
>     Agreed. This is what CDN has done to the Internet. On the other 
> hand, the reverse is true, as well. That is, a centralized mechanism 
> may be subject to distributed control. To illustrate, allow me to 
> recite a small history. I was in charge of coordinating the MIT-LL 
> ground station for the LES8/9 satellite pair nearly half century ago. 
> We had multiple antennas on the roofs with various experiments going 
> on in multiple labs throughout the buildings. They all shared the same 
> set of signals to and from the satellites which meant that we had to 
> have a */centralized mechanism to route / distribute/* such. To do so, 
> I designed/*a coaxial cable switching relay matrix */located at a 
> strategic /*central*/ location with /*identical control and display 
> panels */*/distributed/* at every location where might have the need 
> to configure or monitor the RF signal paths, at any moment.
>
> 2)    " ... Telephone service reliability is enhanced ...":
>
>     Could you elaborate on this? In particular, what "enhanced" the 
> telephone service?
>
> 3)    "... because failure of PSTN and Internet are not perfectly 
> correlated... ":
>
>     When we compare / study two systems, similarity is good enough for 
> learning one from the other. Most of the time, we do not have the 
> luxury of having perfect correlation between them.
>
> 4)    "..  failure of PSTN need not mean failure of ability to make 
> phone call. ....   ":
>
>     By "failure of PSTN", do you mean some parts of the PSTN 
> equipment? If so, you are correct. Otherwise, a system (the entire 
> PSTN) failure would definitely be felt by a subscriber making a phone 
> call. On the other hand, with multiple redundant paths by design 
> serving as backups, strict operation disciplines, plus some tricky 
> user operation instructions, PSTN failures as perceived by subscribers 
> were very rare (true 7- 9's availability versus Internet's promised 5- 
> 9's).
>
> 5)    "  ... may not be known to higher-layer users ...  ":
>
>     I believe what we are looking for is exactly this problem, such as 
> the concentration of Internet routing process to DNS in CG-NAT that 
> eventually exhibited as the concentration of operations to CDN. These 
> results are so general, all the lower level details that you 
> apparently are referring to should have been "smoothed" over by 
> multiple levels of the "integration" process.
>
> Regards,
>
>
> Abe (2024-10-21 18:58 EDT)
>
>
> On 2024-10-21 16:40, William Lehr wrote:
>>
>> distributed and decentralized often mean two different things. A 
>> distributed mechanism may be subject to centralized control.
>>
>> decentralization has important implications for independence wrt 
>> failure modes which impacts reliability. Telephone service 
>> reliability is enhanced because failure of PSTN and Internet are not 
>> perfectly correlated so failure of PSTN need not mean failure of 
>> ability to make phone call.
>>
>> virtualization allows control and function to be separated in 
>> geospace (distributed) that can have implications for performance 
>> (e.g., latency, reliability) that may not be known to higher-layer 
>> users that may seek to exert centralized control over a system that 
>> is distributed, etc.
>>
>> On 10/21/24 10:17 AM, Lixia Zhang wrote:
>>> Hi Abe,
>>>
>>> I think you brought up a set of very interesting, and also very 
>>> important, questions and comments. To me, the top few are:
>>>
>>>   * "distributed or decentralized": it seems worth clarifying
>>>     whether the two words mean the same or different things in the
>>>     DINRG context.
>>>   * the question in Thomas msg showed up again: how much the node
>>>     locations matter.
>>>   * "we find that Internet users have no permanent identity": this
>>>     is a fact, how this fact relates/impacts (de)centralization.
>>>
>>>
>>> To everyone on the list: please share your suggestions about next step.
>>> Dirk and I will try our best to summarize and structure all the 
>>> suggested discussion topics into Dublin agenda.
>>>
>>> Lixia
>>>
>>>> On Oct 20, 2024, at 7:21 AM, Abraham Y. Chen 
>>>> <aychen=[e-mail]> wrote:
>>>>
>>>> Dear Dirk:
>>>>
>>>> 0)    I have been reading the eMail threads and learning the 
>>>> charter of this Group for awhile. I would like to start from trying 
>>>> to understand your first DINRG objective:
>>>>
>>>> 1)    " Investigation of the root causes of Internet centralization 
>>>> ...  ":
>>>>
>>>>     Centralization vs. decentralization in a system could manifest 
>>>> in surprisingly alternating manners depending on which particular 
>>>> perspective that one's investigation is focused upon. For example,
>>>>
>>>>     A.    During the Internet infancy days, the PSTN (Public 
>>>> Switched Telephone Network) was criticized as being too centralized 
>>>> implying that the Internet would be distributed or decentralized. 
>>>> As this Group agrees, the Internet is now far from being 
>>>> decentralized, after all. The reason for this discrepancy is that 
>>>> back then colleagues were looking at the network facility to assume 
>>>> that the operations would have the same characteristics.
>>>>
>>>>     B.    Comparing the operations today, we find that Internet 
>>>> users have no permanent identity, except temporarily assigned by 
>>>> multi-national business conglomerates, thus not able to freely 
>>>> communicate with one another, starting from even a neighbor. This 
>>>> means that the Internet operation today is a centralized one. In 
>>>> contrast, PSTN supports dial-up modems that enable any and every 
>>>> user the freedom / flexibility / independence of contacting anyone 
>>>> around the world anytime for data communication (This was how the 
>>>> Internet initially got popularized, although very slow by today's 
>>>> standards) which is clearly a distributed and decentralized operation.
>>>>
>>>> 2)    The above may sound contradicting that a centralized facility 
>>>> supports decentralized operation while a decentralized one operates 
>>>> centralized. The fact is that the top layer operational behavior of 
>>>> a system determines which way it is from a user's viewpoint. In OSI 
>>>> 7 Layer model, as far as I could understand, we should set 
>>>> distributed/decentralized as the criterion at OSI Layer 7 for every 
>>>> system. Which way the lower Layers appear to be really does not 
>>>> matter, as long as they can eventually support the ultimate goal.
>>>>
>>>> 3)    More specifically, the PSTN core equipment is so centralized 
>>>> that the identity of every subscriber loop is predetermined with a 
>>>> phone number. So, a CPE (Customer Premises Equipment) can be mass 
>>>> produced without the capability to acquire an identification upon 
>>>> plugging into a jack at the end of a subscriber loop. Essentially, 
>>>> the DHCP operation has already been accomplished in a user's mind. 
>>>> In the Internet, however, an IoT has to go through the process of 
>>>> acquiring an IP address (yet temporary) from the network core 
>>>> before it can operate, which means that an user can not operate 
>>>> independently at well.
>>>>
>>>> 4)    The above may be vague, unorganized, or philosophical. This 
>>>> is because I just began to formulate this analytical approach for 
>>>> visualizing the problem at hand. I believe that we do need to 
>>>> exercise our minds from this angle, or similar to be able to set 
>>>> the criteria and priority for qualifying an application or 
>>>> operation. So that, we may have some chance to achieve the goal of 
>>>> "decentralizing the Internet".
>>>>
>>>> 5)    Essentially, I believe that if a user does not have a 
>>>> permanent identity (static IP address) for communication through a 
>>>> system, the operation becomes centralized by having to rely upon a 
>>>> focal facility serving the coordination functions.
>>>>
>>>> I look forward to your thoughts.
>>>>
>>>> Regards,
>>>>
>>>>
>>>> Abe (2024-10-20 10:21 EDT)
>>>>
>>>>
>>>>
>>>> On 2024-10-18 04:41, Dirk Kutscher wrote:
>>>>> Dear all,
>>>>> in Dublin, we are planning to continue our discussion on next 
>>>>> steps for DINRG.
>>>>> To that end, we are soliciting suggestions, interests indications, 
>>>>> and questions here. If you have a suggestion, please feel free to 
>>>>> share it here or by personal e-mail.
>>>>> We will collect everything and then prepare a summary before the 
>>>>> meeting.
>>>>> As a bit of background:
>>>>> As chartered, DINRG has different objectives:
>>>>> • Investigation of the root causes of Internet centralization, and 
>>>>> articulation of the impacts of the market economy, architecture 
>>>>> and protocol designs, as well as government regulations;
>>>>> • Measurement of the Internet centralization and the consequential 
>>>>> societal impacts;
>>>>> • Characterization and assessment of observed Internet centralization;
>>>>> • Development of a common terminology and understanding of 
>>>>> (de-)centralization;
>>>>> • Interaction with the broader research community to explore new 
>>>>> research topics and technical solutions for decentralized system 
>>>>> and application development;
>>>>> • Documentation of the outcome from the above efforts via 
>>>>> different means (e.g., research papers and RFCs) as inputs to the 
>>>>> broader conversation around centralization; and
>>>>> • Facilitation of discussions between researchers, organizations 
>>>>> and individuals involved in Internet standards and regulations.
>>>>> Let us know, which of these objectives should be emphasized in 
>>>>> your view, and whether you have specific interests within these 
>>>>> topics that should be discussed more.
>>>>> Best regards,
>>>>> Dirk and Lixia
>>>>>
>>>>> _______________________________________________
>>>>> Din mailing list -- [e-mail]
>>>>> To unsubscribe send an email to [e-mail]
>>>>>
>>>>
>>>>
>>>> Virus-free.www.avast.com 
>>>> _______________________________________________
>>>> Din mailing list -- [e-mail]
>>>> To unsubscribe send an email to [e-mail]
>>>
>>>
>>> _______________________________________________
>>> Din mailing list [e-mail]
>>> To unsubscribe send an email [e-mail]
>> -- 
>>
>> ==+==+==+==+==+==+==+==+==+==+==+==+==+
>>   Dr. William Lehr
>>   Research Associate
>>   Computer Science and Artificial Intelligence Laboratory (CSAIL)
>>   
>>   MIT Office:
>>     Massachusetts Institute of Technology
>>     32 Vassar Street (32-G532)
>>     Cambridge, MA 02139
>>    
>>     tel:     [phone]
>>     fax:     [phone]
>>
>>   Home Office (preferred):
>>     94 Hubbard street
>>     Concord, MA 01742
>>
>>     cell:    978-618-3775 (preferred)
>>     fixed:   978-287-0525
>>
>>   website:http://csail.mit.edu/~wlehr
>>   email:[e-mail]
>>
>> ==+==+==+==+==+==+==+==+==+==+==+==+==+
>
>
>
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient> 
> 	Virus-free.www.avast.com 
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient> 
>
>
> <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

-- 

==+==+==+==+==+==+==+==+==+==+==+==+==+
  Dr. William Lehr
  Research Associate
  Computer Science and Artificial Intelligence Laboratory (CSAIL)
  
  MIT Office:
    Massachusetts Institute of Technology
    32 Vassar Street (32-G532)
    Cambridge, MA 02139
   
    tel:     [phone]
    fax:     [phone]

  Home Office (preferred):
    94 Hubbard street
    Concord, MA 01742

    cell:    978-618-3775 (preferred)
    fixed:   978-287-0525

  website:http://csail.mit.edu/~wlehr
  email:[e-mail]

==+==+==+==+==+==+==+==+==+==+==+==+==+