[Nsi-wg] NSI Motivation and Context

Jerry Sobieski jerry at nordu.net
Fri Aug 28 12:06:25 CDT 2009


Hi all-

Jeroen vdH and I put together a few points regarding the "conext" and 
"motivation" for NSI.    Please review and comment...

NSI Context

1. Emerging hybrid network services are being deployed in most NRENs.
There is currently no consensus method for automated provisioning of
these services, for asserting policy over the use of these services.
Once provisioned, there also are no common processes for monitoring to
insure service levels or to localize performance problems.
These services are being developed without a comprehensive understanding 
of how these services will be accessed, applied, or provisioned - 
particularly as more dynamic, automated, global, and virtualized 
requirements are emerging.

2. There is a growing portfolio of globally distributed [science]
applications that need predictable and quantifiable network services.
These growing data-centric science work flows require a more integrated
approach that matches network  performance to the acquisition,
transport, computation, storage, and visualization/analysis of the
particular science processes.   The global nature of these
science activities means that these network services will necessarilly
extend across many networks and a myriad of transport technologies to
link collaborating colleagues and facilities.  A common model for
realizing these services on a global basis has not yet emerged despite
ongoing work on control planes for provisioning them.

3. Hand crafted hybrid services for the current early adopters are slow
and manpower intensive.  The development of  automated systems that
allow collaborating research teams to craft application specific 
cyber-infrastructure environments is hobbled by 
the cacophony of locally developed interfaces, services models, and 
technologies.


4. There exists a multitude of industry standards and recommendations
that attempt to address many of the technical issues associated with
multi-domain, multi-layer, dynamic path computation and provisioning,
but these are often confusing or contradictory, and assume traditional
industry service models that limit their usefulness in the increasingly
federated and virtualized global R&E networking and telecommunications
environment.

5. Broad and ubiquitous adoption (or at least availability) of
connection oriented services is a key objective without which maturation
of hybrid services will be difficult.  And adoption will be driven by
the ease and simplicity with which a) the non-expert
cyber-infrastructure user is able to access and leverage these services,
and b) the network service provider(s) can adapt their environments to
meet a common global services model.

NSI Motivators:

1.  A simplified architecture for requesting specific and deterministic
services from the network is necessary to drive adoption.   The NSI,
however, should view the "Connection" as just one possible service for
which a formally defined interface should be defined. The NSI should
anticipate the introduction of other network services and should
therefore be extensible to support a variety of such services.

2. Due to the increasing scope and complexity of co-scheduled global
cyber-infrastructure, automated agents will be necessary and responsible
for specifying and negotiating a complex array of resources - in
particular: network resources.  The NSI should facilitate such automated
service interactions/transactions.

3. As the ability to provision connection oriented services becomes a
more ubiquitous user capability, the traditionally separate roles of
"user' and "network" will merge as application specific
cyber-environments evolve into global private networks, and in turn
offer specialized network (and other CI) services to niche
communities.   An NSI archtitecture that can support both federated
services as well as virtualized services is necessary.

4.  A scalable solution for global service provisioning must recognize
local (domain specific) requirments or situations.  An NSI must be able
to inter-work with such local solutions yet provide a service with
global reachability.









More information about the nsi-wg mailing list