Wednesday, July 20, 2005

The Virtual Von Neumann Architecture

The chain of Automation Quanta (AQ) generated by Internet Systematics is interpreted as the virtual Von Neumann architecture (VVNA) and thus we have an Internet Model.

What is AQ1, AQ2, AQ3,.......

Each AQi represents the mechanization of a previously mannualy executed distributed process by the user agents of the Net which transends its developent stages as

VVNA1, VVNA2, VVNA3,......

Initially IS identified Net's Naming process concerned with the manual maintainance and distribution of the file Hosts.txt [RK] as the first Automation Quantum embedded gradually to the global infrastructure as dictated by RFCxxx - The Domain Name System. Later this was identified as AQ3 (see blog:List of frames generated by IS). The pattern of AQs is repeated with all major advances in the Routing, Addressing, NIR-ing, etc
sub-system components of the Internet System (Internet, ARPANET, WWW are the transitional names the evolving Net)

Let's see how the interpretation works.


Initial development took place as the deployment of the STROWGER SWITCH replacing the manual operators of the telephone system. This form of automation, induced by the STROWGER SWITCH, implemented by elementary devices which influenced the construction of the Von Neumann Architecture. Gradually we arrived at the DIGITAL SWITCH the most basic component of the global Telecom network (PSTN). So we can bind all the activities which lead to the construction of ARPANET under the label Automation Quantum One (AQ1) This creates the infrastructure upon which to overlay ARPANET the first global computer network. All ARPANET computers execute the NCP suite of programs constituting the virtual CPU of the global machine, their filesystems is the virtual Memory and the Telecom lines constitute the virtual BUS. So we can bind all the activities which lead to the construction of ARPANET under the label Automation Quantum One (AQ1).

VVNA1 is made of the folowing components:
  1. the virtual CPU
  2. the virtual MEMORY
  3. the virtual BUS

The nature of AQ2 concerns the automation offered to the users by the application suite of ARPANET - Telnet/FTP/Mail. Can this component be interpeted as

AQ1: manual operator -> strowger switch, digital switch, ncp arpanet node

AQ2: manual telecom communication (voice, text) -> Email, FTP, TELNET, TCP/IP

AQ3: manual Naming -> DNS

AQ4: manual IP routing -> core routing, Backbone, external routing (EGP,RIP,OSPF,BGP, CIDR)

NOTE ---- AQ1.1: manual Telecom -> MPLS/FiberOptic automation

VVNA1: Arpanet
VVNA2: Internet (TCP/IP) - bus=digital telecom
VVNA3: Internet (DNS) - bus= Net w/o automatic naming
VVNA4: Internet (EGP-....-CIDR) - bus= Net w/o automatic routing-manual
VVNA5: Internet (MIME) - bus= Net w/ manual multimedia messaging
VVNA6: Internet (NIR) - bus= Net w/o archie, gopher, www - network services
VVNA7: Internet (QoS) - bus=Net w/ best effort transport

VVNAi+1 =is AQi applied to VVNAi where VVNA1 = ARPANET and AQ1=SS

CONCLUSION: All network advances the last 50 years are modelled by the folowing concepts

(a) the Automation Quantum (AQ)
(b) the virtual von Neumann Architecture (VVNA)
(c) a sequence of AQ(j)
(d) the chain VVNA(i)

The Automation Quantum

Internet Systematics basic result, after observing and experiencing network technology advances as presented by the previous blog: List of Frames generated by IS a common pattern is identified. All advances concern the introduction of a certain level of automation over previously executed processes manually within the net. So unlike artificial intelligence whose advances concern game playing here we have a joint venture of man and computer embarking on a series of advances that we choose to interpret as one enterprise. Advances that initiate with the simplest of communication tasks (operator switching) and reaching the recent innovative activity of the W3C (the tech. body that drives the web) to build the Semantic Web.

Thinking a little deeper we link this concept (The Automation Quantum) with the concept of Metasystem Transition from Cybernetics. It was great surprise that while trying to understand Metasystem Transitions in the computer I discovered a similar concept, that of the Automation Quantum inside the novel entity, we call the Net, computer-like as the phrase 'the net is the computer' indicate.

The concept of Automation Quantum presents a unified interpretation of Net advances that helps our understanding of this evolving complex architecture (a 2nd generation of computerization) in particular as it is undergoing changes continuously. Changes of quantitative nature (e.g Home networking) as well as qualitative advances (e.g video streaming, voice-over-Internet).

List of frames generated by Internet Systematics

Expecting to get a picture of Net evolution from Internet Systematics, I jump right away to the happy-end and present the list of frames that constitute the picture. For each one of the frames I will post an essay, later on, explaining its importance in my analysis. The sequence of frames explains technological evolution at macro scale. One of my initial presentation in 1999 was titled 'Internet under the macroscope', it is in Greek text at the moment but the included diagrams show the main features of Internet Systematics.

There is also intra-frame evolution as in the case of the global telecom network with major changes appearing between decades. Strowger Switch (1950s), MPLS/Fiber Optics (2000) both these major advances concern the automation of the telecom network by replacing the corresponding manual procedures associated with operation and maintainance.

The Strowger Switch

a more humble task compared to the Coloseus of Bletchley Park but both were driven by the same logic, to automate a manual procedure which triggered tremendous innovation forces

Arpanet Packet Switching
Packet switching is the single technological development which accounts

Flag Day 1-1-1983: First and Last manual step (NCP to TCP transition)
Discovering the statement by John Postel as this being the first and last manual act hunted my imagination about his sense of the way Internet was heading to. There the meaning of the phrase 'internet scale' broke clear for me for the first time.

DNS: the automation of the naming process = manual distribution and update of the HOSTS.TXT file
Question : What is the Domain Name System ?
Answer: Automation of a manul (distributed) procedure

Question : What is Automation ?
Answer: The Turing Machine

Question : Can I map onto this concept some aspect of the Computer Network ?
Answer: Difficult ....

Question : To what did The Turing Machine evolve to ?
Answer: The Von Newman architecture (machine)

Question : Can I map onto this concept some aspect of the Computer Network ?
Answer: The total collection of the BIND processors is the virtual CPU. The Zone files is the virtual MEMORY. The virtual BUS is the Net without automatic naming. The I/O to this virtual computer is the RESOLVE DNS client programs.

Autonomous System routing (a series of advances from manual routing )
I read "There is manual routing table maintainance", somewhere (COMER ?) trying to learn how Internet works. Note that at the time did not have the expererience of EBONE and RIPE where I learned about the need to advance the routing technology to care for treating the EBONE core nodes as being on the same LAN.

The MIME multimedia transformation of the Messaging component
MIME is so important that the same mechanism was incorporated by WWW technology.
Its innovator N.S. Borenstein had a bunch of great ideas like 'computational email' which influence my thinking set off by the trade mark 'the Net is the computer'.

The Archie system, the first search engine
while running two student projects titled 'info-hunt' and 'pythia' respectively
to locate, collect and then publish thematic resources on the Internet in 1990 we were overtaken by the Archie system which indexed anonymous ftp sites. It did automatically what we were doing manually and our on-line service 'pythia' never reached the size of Arc hie's indices. But none the less here lies our eureka moments where understanding of net-automation was established.

The WWW series of advances leading to the Semantic Web
It was 1989 that as a member of the WG I was exposed to T.B.Lee's web vision
after some time and whike NSFNET stats were pointing at Gopher traffic with exponential growth I risked my first prediction ' WWW is more scalable system because it has the URL automation (HTML) which is based on DNS automation, hence an overall more powerful system than the manually operated Gopher-root server in Sweden'.

The CIDR change routing engine on flight
Following IETF's discussions on the big-internet List got familiar with the issues such as scaling facing the Net at the time. There I learned about a plan which incorporated the greatest transitional power.

The Automation of the telecom network
Larry Roberts' writings and ideas about the Timeline of Innovations:

Gutenburg's Press > Photography > Radio > TV > Computer > ARPANET

helped my confidence that 'evolution' is a major issue on the road to understanding Internet, I just had to 'decode'its mechanism.

In a white paper about the status of Internet routing technology he refers to 'automating the telecom network' as being the current issue.

The rise of the WITI problem

Looking at Internet History accounts from key people , also at various books about the Internet and its origins, at various articles in eZines and finally at the wikipedia entry about the Internet, surely one gets a good picture about it. For example what Internet is not.

In addition, by participating in a number of projects or activities relating to the growth, usage or development of the Net one gets an experience of the WITI - What Is The Internet - problem. We carried out such activiies in early 90s ( greek text at the moment). Just to give a quick account, it was really interesting times where important components or sub-systems of the Net were emerging such as:
  • Internet in Greece
  • The MIME multimedia Net transformation (got Borenstein's draft from Maria Demou at CERN and built an email system (greek text) with elm and metamail), then found the idea of Computational Email by NSB, a MIME based innovation (early "web computing" idea)
  • The Archie indexing of the FTP-sphere (we were doing it manually as a student project)
  • RIPE control point for the Net (as national representative)
  • BGP II to VI routing transformation (got the experience through EBONE consortium membership)
  • Networked Information component (member of the working group)
The above observations were giving substance to the WITI problem and were shaping the logic of the list of answers to expect.

Is there something else to shed light on the issue ?, a method to construct a picture of the Net ?

I believe that there is and it is Internet Systematics that takes a good shot at the job of constructing this picture.

The meaning of the term "systematics" as in "system" is thus revealed here. Our working hypothesis is that a global computer like system is being advanced continuously (already said by the infamous trademark of Sun Microsystems, of course).

By deploying a method, similar to the idea of F.Baker (ex IETF chair) called "pseudo-area" we observe (in some cases we participate directly) the evolution of Net's components. This allows us a kind of abstraction from the vast amount of information that exists inside the IETF's areas. Of course staying at this informational distance we loose facts and data but hopefully we do not loose the whole picture.

I think, Internet Systematics does something similar to what J.Postel had in mind when he proposed to Rohit Khare to work on the collected technical data about Internet protocols at ISI.EDU (drafts, rfcs, documents) in order to "see" what picture comes out of it. This did not take place according to the latter's statement on the occasion of Postel's orbituary.
I found in Rohit Khare's analysis of Internet problems a similar modality to Internet systematics.

The motivation for setting ourselves such a problem (the WITI problem) was perhaps accidental. On the one hand was the practical aspect of the task what kind of Net to build and the other was the influnce of previous research related to looking for a new computing paradigm (the so-called functional programming paradigm).

Internet Systematics is a language inside which we want to reflect and analyse the properties of the basic elements of the Internet phenomenon. You may also think of it as a kind of theory much like the historical perspectives above but aiming to codify, to create terms and concepts much like a formalism does.

In short, Internet Systematics will generate an evolutionary picture of the Net.