Substrate Resouces
|
View:
New views
20 Messages
—
Rating Filter:
Alert me
|
| < Prev | 1 - 2 | Next > |
 |
Substrate Resouces
by John Jacob-2
::
Rate this Message:
All,
Attached is an excel list of substrate components and resources. One of our near term objectives is to agree on a list of substrate resources and identify the ability to slice or program these as well as classify them as physical, logical or synthetic. Challenges and limitations should also be captured. This list will help the substrate working group in defining the substrate architecture and identifying risks. The attached list is a starting point and is intended to initiate discussion in this area. Please reply to the substrate-wg mailing list with your input and I will maintain a master copy of the spread sheet. Note: This list needs basic input regarding wireless networks. -John --------------------------------------------------------------------- John Jacob Substrate Working Group Systems Engineer BBN Technologies 10 Moulton St Cambridge, MA 02138 (617) 873-8031 _______________________________________________ substrate-wg mailing list substrate-wg@... http://lists.geni.net/mailman/listinfo/substrate-wg |
|
|
|
Re: Substrate Resouces
by Jay Lepreau
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by Masanori Takashima
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by stuart.d.elby
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by John Jacob-2
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by John Jacob-2
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by Keren Bergman
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by Hongwei Zhang
::
Rate this Message:
Please find attached a file that contains the few components/resources that I feel relevant to the "wireless sensor networks" substrate. Best, Hongwei --- http://www.cs.wayne.edu/~hzhang/ John Jacob wrote: All, _______________________________________________ substrate-wg mailing list substrate-wg@... http://lists.geni.net/mailman/listinfo/substrate-wg |
|
|
|
Re: Substrate Resouces
by Drew Perkins
::
Rate this Message:
Some parts of this message have been removed.
Learn more about Nabble's security policy.
Perhaps I’m a bit confused by the word “sliceable”, but my
sense is that this word is interchangeable with “multiplexable”. If so, then
there are several ways of multiplexing a resource. When it comes to CPUs, we
think of time division multiplexing to get virtual CPUs. When it comes to
fibers, I think we can think of not only time division multiplexing (TDM) but
also frequency/wavelength division multiplexing (FDM/WDM) and space division
multiplexing (SDM). A simple example of SDM is that we might partition the
network into multiple pieces for use by multiple experiments. These partitions
may be dynamic based on what experiments are being run at any time. Am I off in
the weeds? Drew
-----Original Message----- Stu, You made some good points and raised good questions. I
would like to know how others on this list feel about this. As a point
of clarification, the selection of
physical/logical/synthetic in the current version was primarily to stimulate discussion. It
is quite likely that several of these assignments are confusing or
incorrect. I will incorporate your input into the master version to be
redistributed after the initial wave of discussions. Thanks, John stuart.d.elby@... wrote: > John, > > I have the same question; I am not sure what the
boundary is between > physical and logical as a result of sliceability.
Building on > Masanori's example, ff a network element has a CPU
running routing > protocol stacks and those processing threads can be
partitioned into > virtual router threads, then I have the basis for
virtual routing > functions (VRFs) which are logical routers built by
'slicing' a physical > CPU. Is this what you were thinking when you labeled
the columns? > > Another observation is that you have labeled 'Ports'
as logical > constructs. I do agree that this is a very
important type of port we > need to consider, but let's not loose sight of the
physical (e.g. 100G) > ports that will be used to connect these platforms
to the underlying > fiber facilities. I suggest that we consider
"line cards" and "switch > ports" physical entities, and "logical
(switch) ports" logical entities > to avoid confusion. The physical binding of a switch
port "a" to a fiber > facility to a switch port "b" would be a
"physical circuit" or > "connection". All other bindings of
logical ports would be "synthetic" > paths as you have already defined. > > -Stu > > -----Original Message----- > From: substrate-wg-bounces@... > [mailto:substrate-wg-bounces@...] On Behalf Of
Masanori Takashima > Sent: Friday, November 02, 2007 12:57 AM > To: John Jacob > Cc: substrate-wg@... > Subject: Re: [substrate-wg] Substrate Resouces > > Hi, John, > > Let me know what do you mean by Sliceable at column
F. > > I guess that Sliceable means that whatever an entry
in column B consists > of a set of finer grained resources. For example,
CPU is Sliceable > because CPU consists of a set of CPU usage, where
CPU usage is a > resource. > > Best regards, > Masanori > > > > _______________________________________________ > substrate-wg mailing list > http://lists.geni.net/mailman/listinfo/substrate-wg > > _______________________________________________ substrate-wg mailing list http://lists.geni.net/mailman/listinfo/substrate-wg _______________________________________________ substrate-wg mailing list substrate-wg@... http://lists.geni.net/mailman/listinfo/substrate-wg |
|
|
|
Re: Substrate Resouces
by Hongwei Zhang
::
Rate this Message:
I think I missed a very important resource for a "wireless sensor node" component --- sensor --- in my last email. Please find attached an updated version of the file with "sensor" included. Best, Hongwei --- http://www.cs.wayne.edu/~hzhang/ Hongwei Zhang wrote:
_______________________________________________ substrate-wg mailing list substrate-wg@... http://lists.geni.net/mailman/listinfo/substrate-wg |
|
|
|
Re: Substrate Resouces
by Peter Steenkiste
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by John Jacob-2
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by John Jacob-2
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by John Jacob-2
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by stuart.d.elby
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by Drew Perkins
::
Rate this Message:
Some parts of this message have been removed.
Learn more about Nabble's security policy.
Stu, I don’t think this is necessarily completely true. Just
as VMWare allows you to virtualize a computer, the same sort of capability can
allow you to virtualize a network (node). There may be some challenges to this
but I don’t think virtualizing the signaling and control plane is particularly
hard. Drew
-----Original Message----- John, >From my point of view, one important attribute of the
substrate is to provide bandwidth 'dial tone' as a sliceable service.
Dynamically requested/allocated bandwidth could range from a full
wavelength, or even block of contiguous wavelengths, down to a 10G
Ethernet connection. To enable this capability, the routing protocol would
need to be aware of the physical topology such as fiber facilities,
physical ports on nodes including end nodes tied to those fibers, and line
card / port resources (e.g., wavelength tuning range, max. port
speed) Imposing such a common signaling and control plane needed
to enable sliceable bandwidth precludes using the substrate to
experiment with new signaling and control protocols, so another attribute of
the substrate would need to be dynamic enablement/disablement of this
feature so that some resources could be allocated for this sort of
experimentation. I think this capability is captured through our requirement
that CPU & memory resources are sliceable. -Stu -----Original Message----- From: substrate-wg-bounces@... [mailto:substrate-wg-bounces@...] On Behalf Of John
Jacob Sent: Monday, November 05, 2007 7:33 AM To: substrate-wg@... Subject: Re: [substrate-wg] Substrate Resouces Keren, This is a good suggestion, but it raises some questions.
Besides bandwidth, what other resources are attributed to the
link-component? Should we include the end-node sites as a link resource?
How would this be different than reserving bandwidth on specific nodes,
e.g. "I want ? Gbps connecting the set of {a1...an} switches. John Keren Bergman wrote: > > John, > > I agree with Stu's comments. Perhaps we can define a
"link" physical > component that includes resources such as 100G. This
would enable > construction of (sub)-networks from the available
components and their > resources - to the extent that these resources can
be partitioned. I would > argue to take out the "network" component
since it's a higher level entity - > it should be possible to build it from the
components we define. The > resources (restoration, protection, QoS, etc.) can
be provided via the > components. > > Best, > > Keren > > -----Original Message----- > From: substrate-wg-bounces@... [mailto:substrate-wg-bounces@...] > On Behalf Of John Jacob > Sent: Friday, November 02, 2007 8:38 AM > To: substrate-wg@... > Subject: Re: [substrate-wg] Substrate Resouces > > Stu, > > You made some good points and raised good questions.
I would like to know > how others on this list feel about this. As a point
of clarification, the > selection of physical/logical/synthetic in the
current version was primarily > to stimulate discussion. It is quite likely that
several of these > assignments are confusing or incorrect. I will
incorporate your input into > the master version to be redistributed after the
initial wave of > discussions. > > Thanks, > John > > > stuart.d.elby@... wrote: > >> John, >> >> I have the same question; I am not sure what the
boundary is between >> physical and logical as a result of
sliceability. Building on >> Masanori's example, ff a network element has a
CPU running routing >> protocol stacks and those processing threads can
be partitioned into >> virtual router threads, then I have the basis
for virtual routing >> functions (VRFs) which are logical routers built
by 'slicing' a >> physical CPU. Is this what you were thinking
when you labeled the columns? >> >> Another observation is that you have labeled
'Ports' as logical >> constructs. I do agree that this is a very
important type of port we >> need to consider, but let's not loose sight of
the physical (e.g. >> 100G) ports that will be used to connect these
platforms to the >> underlying fiber facilities. I suggest that we
consider "line cards" >> and "switch ports" physical entities,
and "logical (switch) ports" >> logical entities to avoid confusion. The
physical binding of a switch >> port "a" to a fiber facility to a
switch port "b" would be a "physical >> circuit" or "connection". All
other bindings of logical ports would be >> > "synthetic" > >> paths as you have already defined. >> >> -Stu >> >> -----Original Message----- >> From: substrate-wg-bounces@... >> [mailto:substrate-wg-bounces@...] On Behalf
Of Masanori Takashima >> Sent: Friday, November 02, 2007 12:57 AM >> To: John Jacob >> Cc: substrate-wg@... >> Subject: Re: [substrate-wg] Substrate Resouces >> >> Hi, John, >> >> Let me know what do you mean by Sliceable at
column F. >> >> I guess that Sliceable means that whatever an
entry in column B >> consists of a set of finer grained resources.
For example, CPU is >> Sliceable because CPU consists of a set of CPU
usage, where CPU usage >> is a resource. >> >> Best regards, >> Masanori >> >> >> >> _______________________________________________ >> substrate-wg mailing list >>
http://lists.geni.net/mailman/listinfo/substrate-wg >> >> >> > > > _______________________________________________ > substrate-wg mailing list > http://lists.geni.net/mailman/listinfo/substrate-wg > > > _______________________________________________ > substrate-wg mailing list > http://lists.geni.net/mailman/listinfo/substrate-wg > > _______________________________________________ substrate-wg mailing list http://lists.geni.net/mailman/listinfo/substrate-wg _______________________________________________ substrate-wg mailing list http://lists.geni.net/mailman/listinfo/substrate-wg _______________________________________________ substrate-wg mailing list substrate-wg@... http://lists.geni.net/mailman/listinfo/substrate-wg |
|
|
|
Re: Substrate Resouces
by stuart.d.elby
::
Rate this Message:
Some parts of this message have been removed.
Learn more about Nabble's security policy.
Drew, I think we are in agreement. My point was
not so much the ease or complexity of virtualizing the control plane; it was the
need to define some underlying assets that include fiber facilities, nodes, and
physical ports on those nodes which would allow any control plane to do the
basic functions of topology discovery, route distribution, etc. To my earlier
point, there needs to be a construct of a physical path node-to-node and/or
end-to-end that can then be partitioned for experimentation. -Stu From: Drew Perkins
[mailto:dperkins@...] Stu, I
don’t think this is necessarily completely true. Just as VMWare allows
you to virtualize a computer, the same sort of capability can allow you to
virtualize a network (node). There may be some challenges to this but I
don’t think virtualizing the signaling and control plane is particularly
hard. Drew
-----Original
Message----- John, >From
my point of view, one important attribute of the substrate is to provide
bandwidth 'dial tone' as a sliceable service. Dynamically requested/allocated
bandwidth could range from a full wavelength, or even
block of contiguous wavelengths, down to a 10G Ethernet connection. To
enable this capability, the routing protocol would need to be aware of
the physical topology such as fiber facilities, physical ports on nodes
including end nodes tied to those fibers, and line card / port resources
(e.g., wavelength tuning range, max. port speed) Imposing
such a common signaling and control plane needed to enable sliceable
bandwidth precludes using the substrate to experiment with new signaling
and control protocols, so another attribute of the substrate would
need to be dynamic enablement/disablement of this feature so that some
resources could be allocated for this sort of experimentation. I think
this capability is captured through our requirement that CPU & memory
resources are sliceable. -Stu
-----Original
Message----- From:
substrate-wg-bounces@... [mailto:substrate-wg-bounces@...]
On Behalf Of John Jacob Sent:
Monday, November 05, 2007 7:33 AM To:
substrate-wg@... Subject:
Re: [substrate-wg] Substrate Resouces Keren, This
is a good suggestion, but it raises some questions. Besides bandwidth,
what other resources are attributed to the link-component? Should
we include the end-node sites as a link resource? How would this be
different than reserving bandwidth on specific nodes, e.g. "I want ? Gbps
connecting the set of {a1...an} switches. John Keren
Bergman wrote: >
>
John, > >
I agree with Stu's comments. Perhaps we can define a "link" physical >
component that includes resources such as 100G. This would enable >
construction of (sub)-networks from the available components and their >
resources - to the extent that these resources can be partitioned. I would >
argue to take out the "network" component since it's a higher level entity
- >
it should be possible to build it from the components we define. The >
resources (restoration, protection, QoS, etc.) can be provided via the >
components. > >
Best, > >
Keren > >
-----Original Message----- >
From: substrate-wg-bounces@... [mailto:substrate-wg-bounces@...] >
On Behalf Of John Jacob >
Sent: Friday, November 02, 2007 8:38 AM >
To: substrate-wg@... >
Subject: Re: [substrate-wg] Substrate Resouces > >
Stu, > >
You made some good points and raised good questions. I would like to know >
how others on this list feel about this. As a point of clarification, the >
selection of physical/logical/synthetic in the current version was primarily >
to stimulate discussion. It is quite likely that several of these >
assignments are confusing or incorrect. I will incorporate your input into >
the master version to be redistributed after the initial wave of >
discussions. > >
Thanks, >
John > > >
stuart.d.elby@... wrote: >
>>
John, >> >>
I have the same question; I am not sure what the boundary is between >>
physical and logical as a result of sliceability. Building on >>
Masanori's example, ff a network element has a CPU running routing >>
protocol stacks and those processing threads can be partitioned into >>
virtual router threads, then I have the basis for virtual routing >>
functions (VRFs) which are logical routers built by 'slicing' a >>
physical CPU. Is this what you were thinking when you labeled the columns? >> >>
Another observation is that you have labeled 'Ports' as logical >>
constructs. I do agree that this is a very important type of port we >>
need to consider, but let's not loose sight of the physical (e.g. >>
100G) ports that will be used to connect these platforms to the >>
underlying fiber facilities. I suggest that we consider "line cards" >>
and "switch ports" physical entities, and "logical (switch)
ports" >>
logical entities to avoid confusion. The physical binding of a switch >>
port "a" to a fiber facility to a switch port "b" would be
a "physical
>>
circuit" or "connection". All other bindings of logical ports
would be >>
>
"synthetic" >
>>
paths as you have already defined. >> >>
-Stu >> >>
-----Original Message----- >>
From: substrate-wg-bounces@... >>
[mailto:substrate-wg-bounces@...] On Behalf Of Masanori Takashima >>
Sent: Friday, November 02, 2007 12:57 AM >>
To: John Jacob >>
Cc: substrate-wg@... >>
Subject: Re: [substrate-wg] Substrate Resouces >> >>
Hi, John, >> >>
Let me know what do you mean by Sliceable at column F. >> >>
I guess that Sliceable means that whatever an entry in column B >>
consists of a set of finer grained resources. For example, CPU is >>
Sliceable because CPU consists of a set of CPU usage, where CPU usage >>
is a resource. >> >>
Best regards, >>
Masanori >> >> >> >>
_______________________________________________ >>
substrate-wg mailing list >>
http://lists.geni.net/mailman/listinfo/substrate-wg >> >>
>>
> > >
_______________________________________________ >
substrate-wg mailing list >
http://lists.geni.net/mailman/listinfo/substrate-wg > > >
_______________________________________________ >
substrate-wg mailing list >
http://lists.geni.net/mailman/listinfo/substrate-wg > >
_______________________________________________ substrate-wg
mailing list http://lists.geni.net/mailman/listinfo/substrate-wg _______________________________________________ substrate-wg
mailing list http://lists.geni.net/mailman/listinfo/substrate-wg _______________________________________________ substrate-wg mailing list substrate-wg@... http://lists.geni.net/mailman/listinfo/substrate-wg |
|
|
|
Re: Substrate Resouces
by Deniz Gurkan
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by John Jacob-2
::
Rate this Message:
|
|
|
|
Re: Substrate Resouces
by Heidi Picher Dempsey-2
::
Rate this Message:
|
Substrate WG Ver0.0.xls (35K) 
| < Prev | 1 - 2 | Next > |
| Free embeddable forum powered by Nabble | Forum Help |
