syslog
27Mar/180

At LSE workshop on data science mar 27/28

Posted by Jon Crowcroft

see
http://www.lse.ac.uk/Statistics/Events/Workshop-on-Data-Science-Theory-and-Practice/Programme-at-a-glance
lots of good speakers! will see if slides will be available - e.g. first speaker's blog
http://inverseprobability.com/2018/02/06/natural-and-artificial-intelligence

23Apr/160

Eurosys 2016

Posted by Jon Crowcroft

some papers caught my eye include:
STRADS: A Distributed Framework for Scheduled Model Parallel Machine Learning.
looks quite clever - not sure how it would work for a bayesian inferencer, but made me think

Increasing Large-Scale Data Center Capacity by Statistical Power
uses MORE servers to reduce power - clever control theory approach - Baidu traces to eval is real, large system

A High Performance File System for Non-Volatile Main Memory.
seems solid

Crayon: Saving Power through Shape and Color Approximation on Next-Generation Displays.
neat, but niche - OLED laptop displays consume less power if you render stuff cleverly - nice bit of human factors driven algorithm design to minimise impact on perceived image quality - gets 56% power saving on tablet with little subjective impact

A Study of Modern Linux API Usage and Compatibility: What to Support When You're Supporting.
Best paper award - nice talk - fun....

BB: Booting Booster for Consumer Electronics with Modern OS.
basically, Samsuung's smart TV boots a lot faster coz they hacked it a lot. (I have one, and replaced an LG with it, and its true:)

TFC: Token Flow Control in Data Center Networks.
is basically isarhythmic flow control (an idea from donald davies 1960s packet switched networking) done right:)

JUGGLER: A Practical Reordering Resilient Network Stack for Datacenters.
uses offload engine and other stuff to do a very solid job of dealing with putting packets in right order for TCP (where out-of-order delivery was caused by load balancers)

Flash Storage Disaggregation.
what it says on the tin

Shared Address Translation Revisited. evil question about reverse page->structure mapping in linux - how to figure out which process to go to with shared stuff...

POSIX Abstractions in Modern Operating Systems: The Old, the New, and the Missing. - hopelessly optimistic but engaging speaker:)

All findable via
http://eurosys16.doc.ic.ac.uk/program/program/

Filed under: Uncategorized No Comments
28Aug/150

Sigcomm 2015

Posted by Jon Crowcroft

A number of us attended ACM Sigcomm 2015conference in London, which was a very well managed affair - hopefully next year (in Brazil will be as good

two things of note here
1/ Heidi Howard won the Student Research Competition
2/ There was an interesting debate around netethics, which George Danezis, et al, blogged

26Nov/130

misled by opportunistic routing…

Posted by Jon Crowcroft

so one of the motivations that led me to think about all the Haggle stuff was the Milgram work. THis seemed to imply, as do pandemics, that a series of chance encounters could, over lots of iterations, provide end-to-end communications reliably.

10 years and 100 papers and algorithms later, it isn't looking so rosy - basically, pretty much every clever scheme has a delivery probability that is quite dismal, even if you set the lifetime of packets/bundles to be many days!

of course, we claim there are other reasons to build such a system (we always did - including reality mining style apps and serendipity)...

but the real problem is misunderstanding the result that Milgram got in his six-degrees work, which made use of the postal network - this isn't an opportunistic network - it is an unbelievably reliable aggregator and schedules delivery between pretty much anywhere on the planet mostly on a near daily basis. ANd the cunning people in the postal service historically provide us with a hint on how to apply this in opportunistic networking..

Indeed, if you go back to the distant past - read about Darwin's voyages on the Beagle (or, perhaps more fun, go see the movie Master and Commander) you'll see that any and all ships going around the world were tasked with the duty of delivering and picking up the post. Without deploying a specific scheduled post boat, instead, all boats going the right way were enlisted as the delivery channel.

Thus, instead of contacts between known friends/acquaintances being required in a successful chain, the weak ties are linked across distances by an early internet - but the early internet was, itself, spatially opportunistic (but "packets" were routed over it in aggregate sacks-worth to the right destination sets).

SO we need to build this (actually, back in Intel research cambridge days, I did propose putting a bluetooth device into the lining of all envelopes and postal bags as I guessed that people walking past the post room in each lab would "drop" and "pickup" their post via the bag, and that would achieve this miraculous shortcut that we've been looking for (without admitting defeat and going back to using the interweb:( )

So we need to data-mine people's use of google maps (i.e. equivalent of lifting their Satnav planned routes) and then use those people who've looked up a place (and maybe mentioned it in a tweet or other social media) to act as ferries (as per georgia tech term) Or mules, or, errr, unwitting post-people....to act as the glue to get us back to the reliable delivery (sure, it still will involve physical travel rather than light-speed photonic networking, but it might get similar delivery properties as the Big-I Internet)

Time to build a simulation and try it based on some trace data....

Building pony express trade routes to strengthen weak ties, and regain the six degrees of separation!!

5Sep/120

Communications and Multimedia Security Workshop

Posted by Jon Crowcroft

Communications and Multimedia Security
University of Kent, Cantervury
Sponsor IFIP
Sep 3-4, 2012

Proceedings are LNCS - will give to CL library if people want to look up any
paper there

Basic conference is fairly good- lots of low level detailed work...mainly securty, but some systems stuff