[ghpn-wg] Fwd: Admela Jukan's presentation

Sat Apr 2 10:52:35 CST 2005

Interesting thread that has fallen off the GHPN reflector for no good
reason. -franco

---------- Forwarded message ----------
From: Bill St.Arnaud <bill.st.arnaud at canarie.ca>
Date: Apr 1, 2005 5:47 PM
Subject: RE: Admela Jukan's presentation
To: Gigi Karmous-Edwards <gigi at mcnc.org>, travos at ieee.org
Cc: chyoun at icu.ac.kr, ebslee at ntu.edu.sg, "Masum Z. Hasan"
<masum at cisco.com>, Leon Gommans <lgommans at science.uva.nl>,
imonga at nortel.com, Admela Jukan <jukan at uiuc.edu>, "Gkarmous at Mcnc. Org"
<gkarmous at mcnc.org>, Cees DeLaat <delaat at science.uva.nl>

Bill,

 I would like to understand more about what you mean by "bringing the
network into the application" - are you referring to vertical
integration from the application down to the physical network
resources? No

A lot of high end applications on HPC machines cannot be decomposed
across multiple machines or grids because they depend on pipeline or
shared memory architectures.  Bringing the network into the machine
that can emulate shared memory and/or pipelining would allow many
existing HPC applications to become more distributed. TCP off load,
RDMA, are some of the underlying technologies that would enable this. 
RDMA, in particular requires consistent network performance and
behaviour – and so is unsuitable for shared IP or fast switched
networks.

 I think this tight integration is what differentiates the Grid
computing from other applications. (this requires a constant feedback
loop between network resources and the applications (or grid
middleware in between))

 Also, 20k Euros for a nailed up 10 Gbps wavelength from what to what?
Some of the e-science applications are interested in host to host Not
necessarily aggregator to aggregator..

Agreed.  But many institutions are building private reseach networks
directly from researcher to researcher and bypassing the aggregator.

Some great presentations on this topic at recent UKERNA meeting . A
number of universities in the UK are building separate private
research networks on their campuses that run in parallel to the normal
aggregated production network. These private research networks for
e-Science, grids etc are linked by dedicated private lightpaths
bypassing the normal wide area networks

Bill

. So dynamic reconfiguration of such pipes will be necessary. Right? I
am referring to applications that choose not to use the routed network
for the various reasons (i.e. Lack of determinism, poor behavior of
TCP over long distances and large data sets, etc.)

 Thanks,
 Gigi

 On 3/31/05 3:13 PM, "Bill St.Arnaud" <bill.st.arnaud at canarie.ca> wrote: 

I enjoyed Admela's presentation on control plane issues.  I think it
is a good summary of most of the issues. However I would suggest there
are some areas that may be worth further exploring:

 (a)    in addition to applications needing to interact with the
network physical layer for large data flows, there are some situations
where it would be advantageous to bring the network into the
application.  This is a lot different than the network being "aware"
of the application.  There is a lot of work going on in the HPC
community to "decompose" large data applications into smaller modules
which then can be relocated anywhere on the network. However in some
cases the application modules may still be on the same physical
machine interconnected by a "virtual" network or pipeline. Extending
HPC pipeline architectures into network pipes would be clearly
advantageous.
 (b)    I remain skeptical about reservation and scheduling of
bandwidth or lightpaths. The cost of wavelengths continues to plummet
– and it is now cheaper to nail up the bandwidth and leave it there
sitting idle, rather than paying the high OPEX costs for scheduling,
reservation, billing etc. For example  I have been informed by
reliable sources that the annual cost of a 10 Gbps wavelength on the
new  Geant network will be in the order of 20K Euros.  You couldn't
hire a graduate student for that price to do the scheduling and
reservation.  The counter argument is that there will be applications
where data transfers are infrequent, and buying nailed up wavelengths,
even at 20k Euros, can't be justified – in that case I say use a
general purpose routed network.  Given that the data transfers are so
infrequent, I suspect the slightly longer delays of using the routed
network can be tolerated.  But I suspect most large data flow
applications will be from well known and often used sources and sinks
– so the need for scheduling and reservation will be very limited

 Bill

 -----Original Message-----
 From: owner-ghpn-wg at ggf.org [mailto:owner-ghpn-wg at ggf.org] On Behalf
Of Franco Travostino
 Sent: Thursday, March 31, 2005 2:44 PM
 To: ghpn-wg at gridforum.org
 Cc: chyoun at icu.ac.kr; ebslee at ntu.edu.sg; Masum Z. Hasan; Leon
Gommans; imonga at nortel.com; Admela Jukan; Gigi Karmous-Edwards; Cees
de Laat
 Subject: [ghpn-wg] Fwd: Seoul material is on-line

 I've been informed that Admela's presentation could not be opened
with PowerPoint. It turns out that the handoff between Admela and me
has altered the file's content  somehow.  I have now replaced the file
in forge.gridforum.org.

 For further reference:

 /cygdrive/D/GGF13 (19) sum Admela*
 59184  2731 Admela Correct File.ppt
 11383  2731 Admela Damaged File.ppt

 -franco

 Date: Wed, 30 Mar 2005 13:08:06 -0500
 To: ghpn-wg at gridforum.org
 From: Franco Travostino <travos at ieee.org>
 Subject: Seoul material is on-line
 Cc: chyoun at icu.ac.kr, ebslee at ntu.edu.sg, "Masum Z. Hasan"
<masum at cisco.com>, Leon Gommans <lgommans at science.uva.nl>, "inder
[BL60:418:EXCH] Monga" <imonga at AMERICASM06.nt.com>, Admela Jukan
<jukan at uiuc.edu>, Gigi Karmous-Edwards <gkarmous at mcnc.org>, Cees de
Laat <delaat at science.uva.nl>

 The whole GHPN production for GGF13 is available at:
 https://forge.gridforum.org/docman2/ViewCategory.php?group_id=53&category_id=941
<https://forge.gridforum.org/docman2/ViewCategory.php?group_id=53&amp;category_id=941>

 We've had a lively meeting (we went 10' past the end of our slot
actually). I hope you will take the time to peruse the minutes and the
material.

 The State of the Drafts that I prepared is thought to be up to date
(alert me if not) ... it also covers a couple of drafts that have been
announced even though they didn't make the GGF13 cutoff date. See
https://forge.gridforum.org/docman2/ViewProperties.php?group_id=53&category_id=941&document_content_id=3603
<https://forge.gridforum.org/docman2/ViewProperties.php?group_id=53&amp;category_id=941&amp;document_content_id=3603>

 The GGF13 program featured a couple of interesting BOFs with strong
network connotation. Kindly enough, both referenced GHPN material.

 One was the Firewall and NAT BOF. The room consensus was that it
should be chartered as a RG.

 The other one was the VPN BOF. 

 On behalf of the GHPN, I invite these groups to use the GHPN
community as a sounding board for their work.  If they don't get the
nod from the GFSG, they can also consider using the GHPN as the
temporary home where to incubate their work further.

 -franco 

 --------------------------------------------
 Gigi Karmous-Edwards
 Principal Scientist
 Advanced Technology Group
 MCNC Grid Computing and Network Services
 RTP , NC, USA
 + 1 919-248-4121
 gigi at mcnc.org
 -------------------------------------------- 

-- 
http://www.francotravostino.name