5 Oct
2020
5 Oct
'20
3:01 a.m.
Someone asked a few weeks ago:
... the Trustees would presumably be completely free to give an
update on the planned relationship with CAIDA (UCSD Network
Telescope), and long-term sustainable plans for AmprGW?
The relationship between ARDC and UCSD's CAIDA research group remains as
it was before Brian Kantor's death. There is a Memorandum of
Understanding (MoU) between UCSD and ARDC that defines this
relationship. The MoU was negotiated between Brian (for ARDC), and the
UCSD management (for CAIDA). In particular, Brian wanted and succeeded
to get this arrangement nailed down before he retired from being on
staff at UCSD. The Network Telescope and the amateurs-to-Internet relay
operate from the same network infrastructure in a lab at UCSD. Both
parties gain from the arrangement. UCSD observes traffic sent to a large
section of unused address space, and has created an analysis environment
to facilitate its sharing of this data with vetted researchers. ARDC
gets a well maintained, high speed interconnect between its users and
the Internet.
Typical amateur tunneled traffic through AmprGW is well under a gigabit
per second, averaging about 30-60 megabits/sec, with bursts 60-90Mbps.
(This traffic occupies twice that bandwidth, since every packet that
comes in, then goes back out through one of hundreds of tunnels; and
vice verse.) Typical non-amateur, Telescope traffic, bursts to 800Mbps
and averages between 500 and 600 megabits/sec.
There are currently no plans to change this arrangement. However, the
main source of funding for the Telescope project expired this year, and
it is not yet clear whether or how the data-sharing (i.e., the expensive)
aspect of the project will continue.
On the plus side, the existing hardware and software that supports ARDC
is all paid for, installed, and running; it would involve work to tear
it down. From the ARDC side, Chris Smith, G1FEF, has full access to
AmprGW from the UK, and continues to maintain it as Brian did, with
intermittent "hands-on" help from a local CAIDA sysadmin. As BDale
recently reminded us, Chris also maintains other ARDC infrastructure such
as the Portal and the website, which run in virtual machines hosted
in various data centers.
If UCSD and CAIDA ever decided to cancel the MoU, shut down the Telescope,
and/or stop collaborating with ARDC, ARDC could move AmprGW to a virtual
machine in a well connected data center anywhere in the world. Now that
ARDC has more than nominal amounts of money, it can afford to pay for
bandwidth and servers. AmprGW remains at UCSD today, partly because
continuing that arrangement was simplest while scrambling to pick up the
pieces after Brian died; and partly to honor the MoU, and Brian's history
there, and to continue enabling Internet research worldwide, since CAIDA
provides access to telescope data to vetted academic researchers.
There are 4 pages of explanation, signatures, etc in the MoU, which is
a public record of the Regents of the University of California, accessible
under the California Public Records Act. Here are the relevent bits:
This agreement is not intended to be legally binding, and instead is
an aspirational document between the parties outlining
responsibilities, and expectations of the parties.
UCSD SHALL:
o Operate network hardware and software to provide colocation services
for the AMPRNet(TM) TCP/IP networks for Amateur Radio on UCSD
infrastructure.
o Agree to safeguard the UCSD equipment and network resources using
best practices for network management.
o Agree to use and comply with best practices for safeguarding data
to mitigate privacy and security concems and to comply with legal
requirements when using the data collected on AMPRnet's network
for research critical to the Center for Applied Internet Analysis
(CAIDA) research group located at the San Diego Super Computer
Center.
COLLABORATOR SHALL:
o Agree to allow UCSD to collect, filter and curate data destined
for the AMPRNet(TM) network for the purposes of network research and
responsible data sharing with the network and security research
communities.
COMMENCEMENT/EXPIRATION DATE. This agreement is executed as of the
date of last signature and is effective through [July 31, 2023] at
which time it will expire unless extended.
The U.S. federal research funding that supports the Telescope is:
https://www.nsf.gov/awardsearch/showAward?AWD_ID=1730661
The proposal that NSF funded is this one:
https://www.caida.org/funding/stardust/
CAIDA's most recent slide deck about the Telescope is:
https://www.caida.org/publications/presentations/2019/stardust_dust/stardus…
The Principal Investigator of the Network Telescope is Alberto Dainotti
<alberto(a)caida.org>rg>. He intends to release a new web site and
documentation for this project by the end of 2020. This will include
a list of research enabled by the telescope (papers, data, analysis
tools).
In the meantime, there is a preliminary Grafana dashboard that shows
that the Network Telescope is seeing (in real time, or from the past).
https://explore.stardust.caida.org/d/ClIeIwOMk/stardust-public-protocols
(It's work in progress! BTW, it uses Keycloak for authentication,
so people can now use github or globus credentials to log in).
Access to the Telescope data is controlled to preserve the privacy of
the users all over the Internet whose (typically malware-contaminated)
sites originated the packets. Researchers who use the data must sign a
contract agreeing to maintain that privacy. Note that none of the data
in this Network Telescope is the traffic of authorized amateur users.
All that traffic is filtered out before it is recorded for researchers
by the Telescope.
We are happy to take questions or feedback on this list or at the
community meeting next week.
John Gilmore, W0GNU
Board member and Secretary, ARDC