Team 1 – High Level Document Review
Document Name and Location
Root Scaling Study: Description of the DNS Root Scaling Model by TNO
http://www.icann.org/en/committees/dns-root/root-scaling-model-description-29sep09-en.pdf
High Level Summary
The purpose of this document is to describe the root DNS scalability model, as a basis for
explanation of the quantitative root scalability model for the RSST and potentially other interested
audience as well as development of the quantitative simulation model by TNO.
The scope of the quantitative scalability model is the provisioning and publication subsystems of the
root server system. The art of modeling consists of including all relevant parameters and relations
between them that capture the dynamic behavior of the modeled system, while omitting all other
aspects that are not relevant to the goal of the model based analysis.
Key points relating to Security, Stability and Resilience
This report describes the quantitative model and its software implementation that was developed as
part of the results of the RSST. In combination with the broader results from the RSST the
quantitative model will allow the community to anticipate on the consequences of combinations of
changes to the root system; identify “early warning signs” of system stress; and plan ahead for the
mitigations that may be necessary to keep the system running smoothly if and when those signs
appear.
Key SSR sections to note are:
The validation tests with the ExtendSim implementation of this baseline model illustrate that the
model predictions concerning lead times in the change request provisioning process and the zone
file loading process are plausible:

the order of magnitude of the model output (e.g. change request lead times) are within the
range of values that are encountered in practice and

the ‘breaking points’ are roughly around the input parameter values where they are expected
to be.
The modeling exercise itself has shown that there remain questions regarding several details of its
current operation and that it was quite hard to find quantitative data as input for the model.
In case the desire rises to give more accurate boundaries for the scalability of the root, then the
model should be taken one step further. The actions needed to do so are:

Collect (more accurate) quantitative data as model input. This action can typically be
combined with the general recommendation to start monitoring the root system in order to
get “early warning signs” of system stress.

During model development we chose some scope regarding the purpose of the model. Once
the community starts using the model specific feedback regarding the use of the model should
be gathered.

The model should be fine-tuned and validated more extensively using this collected
quantitative data and the more specific intended use of the model.

After this fine-tuning step the model can be applied to conduct a root scaling analysis in order
to give more accurate boundaries for the scalability of the root.
Any Key Areas/Information missing from the Document?
As noted in the document, the model lacks the error validation, which is of equal importance as the
zero-failure validation. However, the error validation issue is addressed not only by the availability
of quantitative data but also by the in-depth analysis of the error occurrence model.
Conclusion
This is a report by TNO detailing the DNS Root Scaling Model. The no error validation illustrates the
plausibility of the model description. But more efforts are still needed to incorporate the error cases
into the model both analytically and quantitatively.
Scoring Framework
1
V. Low
How relevant is this document to Security?
How relevant is this document to Stability?
How relevant is this document to Resiliency?
How useful is this document to the SSR-RT2
overall?
How accurate/factual is the document?
How controversial is the document?
How detailed is the document?
How complete is the document? (Is it missing any
pertinent information)?
How timely is the document? (is the document up
to date or out of date)
2
Low
3
Med
4
High
X
X
X
X
5
Critical
X
X
X
X
X