COOK Report for July - August 2013

Building High QoE Networks without Increased CapEx

How it is Done

What follows depends on your having read the ten thousand  words of the May June 2013 COOK Report.  There you will be introduced to the concept of treating networking as a science rather than as a craft.  Ken Miller has been of enormous help in understanding the mathmatical and engineering details as well as the concepts involved in the way in which IP Networks can be engineered to deliver performance while enabling the postponement of equipment upgrades.  The content of this issue is highly dependent onhaving read and internallized the previuous issue which is an introduction to how PNSol uses Neil Davies tools to enhance the performance of the networks of his clients.

Delta Q or quality attenuation is the key concept that underlies the work of PNSol. However the use of "delta" as the means of expressing quality attenuation Ken explains the basic concept of “delta” in engineering and does it so well that he should get everyone’s attention.  He then goes on to anthropomorphize the PNSol approach to networks as a science.  He tells a story in a few hundred words that summarizes the 50 page text of this issue.


What is ΔQ?

These two characters are the symbol of an emerging technology that can be used to drive more efficiency out of congested IP networks.  Like many “new” ideas, it struggles with acceptance and suffers the curse of the current predominant concepts of network quality.  Don’t be fooled into believing it is just another QoS technology, ΔQ will not only affect the IP networks of today, but it will likely shape all network constructs into the future.
For the non-math geek and non-network geek, you could hardly find a more obscure title for this emerging concept.  For the digerati and closet math junky, I don’t want to belabor the obvious, but a few moments to tease out the meaning in this modest abbreviation may go a long way to make this concept accessible to everyone.

For the non-math geek, the Δ character, aka “delta”, is used in calculus to signify a variable unit of change (not a Greek fraternity).  This term carries with it the conceptual weight of calculus and engineering.  The symbol Δ connotes calculated results, not some esoteric emotional state.  A delta is something that is precisely measured and wielded in a way that can crack the secrets of the physical universe; quite a bit for one little symbol to carry around.

The Q, in this case, is for Quality; not to be confused with “premium” and “better than average”.  Although this term has been used before by network engineers to describe “quality of service”, the quality in ΔQ isn’t just talking about a getting premium quality; it’s talking about “perfect quality”.

Why is this subtlety important?  Because this difference is the conceptual pivot of ΔQ.  It is the complete reversal of the concept of trying to engineer more quality into communications.  It is about engineering less degradation.  Doesn’t seem like a big shift.

If you listen to the cerebral and philosophical evangelism by Martin Geddes around this technology, you can get a glimpse of what the Q is about.  Martin’s use of “translocation” to describe communications networks is talking about this kind of “Q”.  Martin uses translocation to mean the instantaneous movement of information from one location to another.  Translocation is perfect communications.

ΔQ is describing everything less than instantaneous and error free communication.  Put another way, all communications is less than perfect.  ΔQ is about describing the attributes of the network which degrade this perfection.  This is the opposite of other discussions about network quality which focus on what can be done to increase the quality of communication (faster switches, bigger buffers, more circuits).  These quality discussions are about managing up from what we have, ΔQ manages down from perfection by managing/engineering degradation of all network flows.

The Story of the Island Traffic or Ken Miller's EWxplanation of a Poly service Network

With Hofstadter and Dennett as my idols, I cant help but attempt to translate ΔQ into accessible concepts.  If they can do it with philosophy and cognition, surely ΔQ couldn’t be too difficult.  The following is my attempt to construct a metaphor to explain the fundamentals of ΔQ.  Like all metaphors it is flawed and can be misleading.  I am certain that there are better metaphors and this explanation can be refined over time, but I believe it is a good enough start to allow non-network engineers a glimpse at the paradigm shift that is called ΔQ.

Imagine a Toll Bridge connecting a private island to the main land.  The island, is large enough to have large cities, industries, and resident population.  For years the toll bridge served the island community well, but as life moved into the 20th century, the residents of the island began experiencing random and massive traffic jams crossing the toll bridge to the mainland.

As the traffic jams escalated, some residents began moving to a neighboring island with a brand new state of the art toll bridge. The Mayor of Island Beautiful, impacted by the negative polling numbers and genuine concern for the island’s long term economic future, began to seek out solutions.

After a rigorous search, the Mayor had two proposals on his desk; one from the same engineering firm that had built the toll bridge for the nearby island and one from a previously unheard of consulting firm.  They had almost thrown out the second proposal due to the fact that it didn’t make any sense to the island engineers, but the firm’s references had checked out.

The only reason the Mayor had not already funded a new bridge was that the island nearby had begun to experience their own traffic jams.  In addition, the second firm’s proposal wouldn’t require a year of new construction, which would further impact his polling numbers in this election year, so he signed off on the consulting proposal, thinking that if it failed, he could always start new construction in a few months later.

Within a month, a young engineer stood before the City Council meeting to explain the results of the engineering study and the plan to fix the traffic jams.  The engineer explained that all of the impacts of the island toll bridge traffic could be represented by three different island natives, all who were present for the meeting.  After the representatives had a chance to explain their problems, the engineer would explain the plan.

The first and most vocal resident was the owner of Zippy’s Pizza.  She explained that the traffic jams kept her from delivering pizza to the mainland in a timely fashion.  When the traffic jams impacted her drivers, all the pizza was cold and customers would not pay.  There was no way she could continue to deliver pizza to the mainland with these random traffic jams, it just wasn’t worth the risk of losing all that pizza.

The second representative was from Phil’s Dirt Transportation Company.  Phil patiently explained that when his massive trucks sat idle in the traffic jam, he had to pay for all the fuel and driver time, which drove up his cost of dirt to the point that he could no longer compete with the mainland dirt companies.  His issue was that he needed to keep his cost for delivered dirt as low as possible; he just couldn’t afford to have trucks waiting in line.

The final representative was General Webb.   The General explained in vague terms how he hated sitting in traffic and how it seemed that other islands had better bridges.  BUT, he was adamant, that there was no way he was paying more to fix the problem that was clearly caused by Zippy and Phil.  Although the General’s complaints were too vague to help much, the Mayor and the Council knew that he represented a major voting block on the island.

As the discussion began to go in circles, the young engineer finally stepped up and asked to begin his presentation.  He began by stating that the island didn’t need a new bridge if everyone was willing to make a few compromises.  Although everyone immediately objected and argued that no one should give up anything, the discussion eventually came back to what exactly would they have to give up.  The engineer assured them that they would only be asked to make compromises when the bridge began to back up.

The engineer described that the island had a bridge with multiple toll booths.  All traffic destined for the mainland took the first available toll booth or began lining up in the shortest line.  After observing their traffic, he said that 1) All the traffic was randomly arriving at the bridge 2) most of the time, the delays crossing the bridge weren’t a problem for anyone 3) a couple times of the day, at unpredictable times, the traffic would back up and cause problems for everyone.

The solution, he described, was to install three symbol lights at the top of each toll booth.  Zippy and like companies would have to register in advance for a yellow triangle pass.  Phil and like companies would register in advance for a brown square pass.  General and like traffic wouldn’t have to register, but would only be allowed to use green circle toll booths.  As long as there was at least one open toll booth, all toll booths would show a green circle, yellow triangle, and a brown box.  This would mean that any type of traffic could go through any toll booth.  Everyone was nodding in agreement, this is how things were now and it worked.  So far no one had to give up anything.

If all the toll booths were full, one toll both would change display to a yellow triangle only.  This toll lane would then be dedicated to Zippy and companies like Zippy so that no other traffic would be allowed through this toll booth.  Zippy drivers could also use all the other toll booths.  Additionally, if Zippy’s drivers were still being delayed, they would continue to dedicate toll booths, until a pre-determined maximum number of booths were reached.  After this, Zippy type traffic would be delayed, just like everyone else.  The owner of Zippys began to smile.  The General and Phil began to shift in their seats but they knew that there wasn’t much Zippy traffic going over the bridge compared to the rest of the traffic.  The engineer quickly moved on to the next step.

He continued; if all lanes were still full, the booths would begin turning off brown squares until there was an average wait time for the green/General traffic (even through Zippy traffic would be sometimes mixed in with the green traffic).  Phil exploded at this time; objecting to being marginalized and potentially unable to transport his dirt.

The engineer pointed out that they would always keep at least a couple toll booths open for brown square traffic (but this booth would always still have the green and yellow signs lit as well).  The engineer then explained to Phil that if there was more than one truck waiting, they would send all other trucks back to Phil where he could park them and put the drivers to work doing something else.

Although still angry, Phil began to consider that no one was “waiting” on the dirt delivery on the mainland and that this scheme would prevent him from continuing to send out more trucks if he had them returning and parking at the dirt mine.  This would keep his costs down.  Plus he realized, most of the time his trucks were going through without delays, it was just the jams that were costing him money.  Overall this wouldn’t impact the volume of dirt he should transport, it would just delay it a bit, but it would eliminate the costs caused by the traffic jams.  Phil slowly agreed to the scheme, but not without being able to change back if things weren’t working well.

The General summarized to make sure he understood.  Zippy traffic could always use all booths, but in a jam would get dedicated booths.  Phil could only use booths with brown squares and was ok slowing down during traffic jams.  He would always have access to lanes and would only be restricted by Zippy lanes.

The General sat quietly thinking until he brought up the fact that even with this scheme he could still run into traffic jams.  The engineer nodded and indicated that this solution did not prevent traffic jams; if there was simply more traffic than the bridge would handle, the General would get caught in jams.  This method is about adjusting the traffic to allow the most Zippy, Dirty, and General traffic over the bridge as possible before they had to upgrade to a new bridge.  Consider that without this method, Zippy and Phil would stop doing business with the mainland long before the bridge capacity was at it’s maximum.  By allowing Zippy and Phil to prosper the taxes would lower the General’s costs for a new bridge down the road.

Zippy’s owner and Phil nodded slowly in agreement and everyone in the room began to understand that no one got everything they wanted, but this solution would get them all further down the road.



Executive Summary                                     p. 3

Do it Ourselves Networks
A Field Trip Approach to Grass Roots Networks and Call to Approach
Data Translocation as Science Rather than Craft                     p. 4
Overcoming Four Decades of Perceived Learning about
Computer Net-works                                          p. 5
Note to Readers                                                 p. 6

Introduction by Ken Miller
What is ΔQ?                                                p. 8
The Story of the Island Traffic                               p. 9

The Derivation of Neil Daviesʼ Calculus of Networks
What Predictable Network Solutions Does and Why It Is Important    p. 13
So to start with ∆Q of Data Transport                            p. 19
Impairment of Quality                                        p. 22
April 8 Ken Miller                                            p. 24
April 9 ∆Q expressed in G,S and V terms                        p. 26
Broader Questions                                            p. 32
April 10: No Need for a Central Agent                            p. 33
April 26 Teleconference Discussion on Multiple Nodes                        p. 38

The Discussion                                                p. 39
Large Scale Control Mechanism Needed?                        p. 42
Quality Attenuation Budgets and Failure Modes                p. 50
Advantage of Knowing When You Must Upgrade                p. 54
Traffic and Backhaul                                            p. 55
Economic Impact of Reliable Network Infrastructure              p. 57
The Disconnect of Bandwidth Versus Services                p. 60

Conclusion: Ken Millerʼs Summary for the Citizen                                                    p.  61
COOK Report’s Final Thoughts                                p.  65