lmbgp - Tumblr blog

[lmbgp]

“the morning paper” covers our Research for Practice recommendations

Wow! "the morning paper” is a fabulous blog and I’m super stoked to read Adrian Coyler’s take on the papers we recommended. Check out his posts on:

E2: A framework for NFV applications

mcTLS: Enabling secure in-network functionality in TLS

BlindBox: Deep Packet Inspection over Encrypted Traffic

[lmbgp]

Now, about that thesis... this is all I have. The title page.

[lmbgp]

Blog / life update after many months of interviewing: 

I will be joining the faculty in the computer science department at Carnegie Mellon in July 2017. 

My spouseman will also be joining the CMU faculty. We’re pleased as punch.

[lmbgp]

The abstract to Murphy et al’s HotNets Paper is delightful.

[lmbgp]

Reflections on SSN2015

Last week I had the honor of being invited to speak at the Spring School on Networks in Santiago, Chile. Next year is a big year for Latin America in the networking research community: SIGCOMM will be held in August in Brasil, and an IETF meeting will be held in April in Argentina. These major international conferences are primarily US and Europe based. Hence, many students rarely have the opportunity to engage with the international research community with conferences being held in very far away and expensive-to-reach locales.

The Spring School on Networks brought together students from Chile, Uruguay, Mexico, Colombia, and Spain to bring together the Spanish-speaking networking community, introduce students to some big ideas in networking research, and encourage them to submit to and participate in the upcoming international events coming to South America. I was so stoked to be a part of it! Here’s what I learned, both in terms of research/technical content, and meta-research/research-community lessons.

Muchas gracias to Javier Bustos, who organized and invited me!

"Output commit pasa caleta" - Justine Sherry Presentando en la @SpringSchoolNet de las Jornadas Chilenas de la Compu pic.twitter.com/q0M7svXXmh

— Omar Miranda (@ommirandap)

November 9, 2015

Research/Tech

• Internet Access.  The workshop started off with a keynote from José Piquer, sender of the first email in Chile, who was part of bringing Internet access to Chile, involved in managing .cl, etc. He mentioned two anecdotes I loved about why he felt it so important to bring Internet access to Chile: one that they were thinking about at the time they established connectivity and one he realized more recently.

The first benefit of access was freedom of information. This sounds cliché, especially to American audiences. Yes! It’s the Information superhighway! Libraries are dead! World news! We all know this!

But actually, we don’t quite understand it in the way that Professor Piquer does. At the time they established their first international links in 1987, the Pinochet dictatorship was still in power, mounting huge pressures on the press to report favorably about the government. Censorship was the norm. Just five years earlier, they had watched their neighbors in Argentina essentially lose a war against the UK over the Malvinas/Falklands islands – while the press all the while reported they were winning.  Internet access changed that. Five years later, Chile was connected to the NSFNet, and unrestricted information access was there to stay.

The second access issue – the one he noted recently -- was the role of networking in emergencies. Last year, Chile suffered a huge earthquake, in which many people injured or died. Coordinating disaster relief on a huge scale was difficult, but, as Prof. Piquer noted, “WhatsApp saved the day.”

• The IETF. My favorite talk at the whole workshop was not strictly technical – it was a talk about the workings of the IETF given by Sandra Céspedes. I have never seriously been involved with IETF anything before, although moving forward there are things in my research that I would like to standardize and so I knew I needed to learn how things work! Just my luck, Professor Céspedes gave a thorough run-down of the organizational structure and standardization process with the IETF.

The thing that surprised me the most was that I did not realize previously how open the IETF is – to participate in a working group, all one has to do, really, is to sign up for a mailing list and just dive in. That’s it. No credentials or vetting required. Another thing I loved learning about was the different types of standards: proposed standards, draft standards, and Internet standards. To become an Internet standard, one must have multiple working and interoperable implementations of the proposed protocol. It’s a nice and clean way of defining what is a “real” standard (as in, something people use) vs what is a not really a “standard” at all (it’s just something IETF thinks is a good idea, but doesn’t seem to have broad adoption).

Another thing I noticed, “meta” the talk – the IETF has been a fabulous venue for Professor Céspedes to have international impact. She did her PhD at Waterloo, a super well known North American university, but then moved to ICESI in Colombia, followed by University of Chile – two less well known, less well-connected places. By tapping in to the IETF, she can still push forward her agenda internationally, despite being far away from many of the folks she’d like to interact with. Which is awesome!

• Vehicular networks are going to be a thing. The second thing Prof. Céspedes convinced me of (she gave two talks) was that vehicular networks are going to be a real and important thing. I had always looked at vehicular networking papers as essentially toys that researchers liked to play with because they’re fun and have different properties (lots of mobility and motion especially) than normal networks we work with. And it’s fine to work on things just because they’re technically fun – but disappointing when there’s no impact to be had. I was 100% wrong in the impression that there’s no impact to be had.

Two facts: (1) By 2020 all new US and German cars must support 802.11p wireless networking in order to support new emergency features (e.g., a car accident automatically triggers a call to 911). (2) Google’s self driving cars and Tesla’s autopilot exist, and in general cars are becoming increasingly computerized.

I will be watching this space much more seriously as we go forward in time.

• Latin America wants a major datacenter. This point came up during a talk given by Goranka Bjedov from Facebook. None of the major Internet players – Facebook, Amazon, nor Google – have a datacenter in Latin America. They do make aggressive use of CDNs to improve performance, but the best access to their features in Latin America will come when they actually host datacenters closer by. Furthermore, Latin American countries don’t want to continue to rely in US-based infrastructure, given the fact that US government officials at this point routinely access international data that is routed through US territory. Brazil is aiming to send all their data to Europe instead.

One woman in the audience – I’m not sure where she was from – mentioned that the Chilean government is sinking fiber in to the southern tip of Chile to try to entice these companies to open a datacenter there: temperatures are cold, and with this new fiber there will be fast Internet connectivity. And of course, there is a huge market in Latin American that wants access to services from a more local cloud.

Meta-Research/Community

• Language is HARD! This was the first conference I ever attended in not-English. Many of my friends know I’m taking Portuguese to be able to communicate with my future in-laws, but actually I’ve been studying Spanish for much much longer and can function at roughly a B2 level on the CEF scale. Basically what this means is that I can read books without a dictionary (although I’ll miss some words) and watch TV without subtitles (although strong accents are still difficult to follow).

Whatever, attending that conference was HARD. It turns out I can’t write and take notes at the same time that I listen to a talk! Also, José Piquer has the strongest Chilean accent I’ve ever heard, which means all words are one syllable shorter! Also, slang! Also, it’s a bit embarrassing to give a professional talk knowing that I make tons of grammatical mistakes!

The most difficult: it was near impossible to convey ideas that are subtle or intricate. If my brain is a dual-core processor, I think I usually have access to both cores to think about how to explain things or try to understand things. But when discussing technical content in Spanish, I was down to one core – all of my ideas came together just a bit slower. It’s already hard to explain in English, e.g. output commit from FTMB, or what makes middlebox deployments complex.

Overall, this has deepened my empathy for international students who are forced to do this every damn day in their research careers. I always knew it was hard, but had never tried it myself. Nonetheless, I had one social benefit most international students still don’t have: the bar is low for Americans internationally. No one expects us to learn a new language, and people were so kind and impressed that I had done so. International students have to deal with Americans who judge them for having strong accents rather than congratulate them for learning something new. And that’s super unfair.

• Geographic outreach. There is less funding overall for research in Latin America. The workshop itself was funded by SIGCOMM and the Internet Society and I think it was a good investment. Our communities are incredibly US, sometimes Euro, centric, with the occasional paper from China (usually MSR or Tsinghua).  We’re missing out on a lot of bright young folks – especially students! – if we don’t take steps to pull them in.

I think hosting more location-oriented workshops like this is a good way to try to connect the broader geographic world to our research community and not to lose out on great minds just because we don’t connect to them. I hope the students who attended the workshop will feel ready when they get to SIGCOMM or the IETF meeting, and be confident to contribute to the technical conversations going on there.

[lmbgp]

Webbernews

CAIDA releases the August 2015 Internet Topology Datakit. I’ve used this for research in the past -- it’s incredibly useful to have  router-level graph of the Internet for all sorts of things!

Amazon will open a London datacenter. I mean “datacentre.”

Comcast is extending usage caps. My house could be in trouble...

Running a telco is getting increasingly complex!

Did you know that Colorado has a ban on municipalities providing Internet unless residents vote in a referendum? This is a stupid obstacle to making Internet access more widespread. 15% of American residents still do not have Internet access.

An editorial on zero rating, billing, and pricing policies on the Internet. 

PRIVACY ALERT: In the UK, your browsing history must be stored for a year by your ISP.

What data are your apps leaking about you? (We saw some of this in the past with Meddle in some lovely research done by then-undergrad Amy Tang).

[lmbgp]

Commernetterwebs

Here are some things I found interesting today:

Russian Navy’s “aggressive” activity near underwater cables worries US.

NY suspects Verizon, TWC, and Cablevision of slowing Internet traffic.

Facebook tries to bring Internet basics to India (Again!) Facebook seems to be doubling down on extending Internet access everywhere. Check out this Berkeley startup whose founders just moved to FB.

[lmbgp]

Shared for the delightful term “middlebox terrorists”

Copy/pasted directly from the e2e archives; thx to Fernando. 

[e2e] purpose of pseudo header in TCP checksum

David P. Reed

Tue Feb 15 04:39:39 PST 2005

As I was there (in 1976, when we split TCP into IP, TCP, and other protocols, such as UDP) for the decision to separate the checksums and to create a pseudo-header, here is the rationale, which is highly relevant. TCP (and UDP) are end-to-end protocols.   In particular, the TCP checksum is "end-to-end".   It is a "private matter" between end points implementing the TCP layer, guaranteeing end-to-end reliability, not hop-by-hop reliability. IP is a wrapper for TCP, which instructs the transport layer (the gateways and routers) where the packet is to be transported, how big it is, and how it may be fragmented in the process of delivery.. The Source Address, Destination address, length, etc. are part of the meaning of the TCP frame - in that the end point machines use that information in the TCP application. Thus the function of SA, DA, etc. are "shared" because they are meaningful to both layers (IP and TCP).   Rather than include the same information twice in the packet format, the concept of a "virtual header" was invented to encapsulate the idea that IP is not allowed to change the SA and DA because they are meaningful. Further, in the case of end-to-end encryption (in 1976 we had a complete design by Steven T. Kent, my office mate, which was blocked by NSA from being deployed) it is essential that all end-to-end meaning be protected.   The plan was to leave the SA and DA in the clear, but encrypt the rest of the TCP payload, including the checksum.  This would protect against a man-n-the-middle attack that delivered valid packets with an incorrect source address, etc. (yes, to be truly reliable, we would have had to use a DSA instead of the current checksum). This was a careful design decision, wrecked irrevocably by the terrorists who invented NAT (which doesn't allow end--to-end encryption, because NAT is inherently a "man-in-the-middle" attack!). The rise of the middleboxen have now so thoroughly corrupted the Internet protocol design that it's not surprising that the original designs are difficult to decode.   If we actually had end-to-end encrypted TCP (now impossible because of the NATs) we would have a much more secure and safe Internet, while preserving its open character.   Instead we have a maze of twisty, disconnected passages, vulnerable to a zillion hackers.

[lmbgp]

So many empty promises.

I’m so sick of these. University chancellor says:

“We must do everything in our power to create the conditions necessary for quick and confidential reporting of suspected violations of our rules and standards of conduct.” 

The problem here was not that nobody reported. Four women reported. The problem was that you didn’t *do anything about it.* What good is more reporting if there are no results?

Also, from my own department:

“We would like to assure you that the EECS Department has a zero-tolerance policy; any harassment or discriminatory behavior that is brought to our attention is reported promptly to university authorities.”

So... the zero-tolerance policy here means that you will always report to the university, who have basically demonstrated that their policy is “we tolerate this?”

Here is the statement I want: Being found guilty of sexual harassment is grounds for termination of employment at this university, regardless of whether or not the perpetrator has tenure, how many grants they bring in, or how many sparkly awards they have.

[lmbgp]

Chancellor Dirks’ List of Excuses.

“Once the investigation in to Professor Marcy’s conduct concluded, Berkeley’s leadership considered disciplinary options, but did not have the authority, as per University of California policy, to unilaterally impose any disciplinary sanctions. Discipline of a faculty member would include an additional faculty-led investigation, and a full hearing before a faculty committee, is a lengthy and uncertain one. In addition, the standards of evidence that would be used in that process are higher than those that are applied by the Office for the Prevention of Harassment and Discrimination(OPHD) in the course of its investigations. The faculty-led process would also be subject to a three-year statute of limitations.“

[lmbgp]

Dot Webbernews.

Do we need gigabit networks for everyone?

North America ran out of IPv4 address last week. So long and thanks for all the Internets.

Microsoft announces new features for Azure.  Two things that are rad here:

New regions in India. In APLOMB, we found that our network processing service worked great (in terms of performance) within the continental US -- move beyond that, and performance drops because users are far away from their nearest datacenter. This is (a) a sign of better service for people in more remote regions, (b) a sign that MSFT is starting to prioritize services for non Euro/US customers.

They also support Mesos, which was invented here at Berkeley by some cool kids in the RADLab.

Let’s all say “cyberthreats” as fast as we can on repeat. Also, a US-China “cybertreaty”? An editorial here.

Google to provide free Wi-Fi in railway stations in India. (More targeting India...)

Cookie handling can break HTTPS :(

Cisco planning to partner with a Chinese company to get better access to the “local market”. Better title, “Cisco afraid of Huawei.”

Only sort of Internet related: Is Blackberry finally going to bring a competitive hardware keyword Android device to market? I miss hardware keyboards deeply and am waiting their return...

Also only sort of Internet related: Pebble is about to release the first smartwatch that is not so ugly that I can’t bear to wear it. NOTE EVERYONE WHO MAKES SMARTWATCHES: LOOKS MATTER. On my Christmas list, please, Santa.

Check out this brilliant MIT Professor, and Berkeley alumna, who rocks.

[lmbgp]

Interview on the state of broadband access worldwide as surveyed by the UN. Still only half of the people in the world have access to *any* Internet connectivity at all. 

Read the full report here.

[lmbgp]

When I was in the fifth grade, my Odyssey of the Mind team won our district competition and we got a plaque. One plaque, six students. As far as I know, it’s sitting in the office at my elementary school still.

Props to SIGCOMM for sending one of these to all eleven authors! This is *way* better than fifth grade.

Edit. Cannot figure out how to get tumblr to not make that photo so ginormous.

[lmbgp]

Dying of laughter. Security is doomed. The fabulous James Mickens, folks.

[lmbgp]

Rollback Recovery for Middleboxes. SIGCOMM 2015 talk and Best Student Paper Award.

[lmbgp]

Deep Packet Inspection over Encrypted Traffic! My SIGCOMM 2015 talk on our system BlindBox.

[lmbgp]

A gem on teaching in “A Prescription for Black Doctors” in the NYT today.

Francis believed that Xavier should not follow the example of most pre-med programs — ‘‘Look to your left, look to your right; only one of you will still be here at the end’’ — which work to weed out students. To him, that model squandered the talent of far too many students, especially black ones. Instead of compelling students to compete against one another, he said, it made much more sense, both morally and practically, to encourage better-­prepared students to help their classmates who weren’t as fortunate to catch up.

...

Students would stay up until the wee hours of the morning helping one another. ‘‘You have almost a hundred kids asking questions, discussing the material,’’ Johnson said. ‘‘To see the material broken down that way was just amazing. And if you didn’t get it, they’d explain it again. And if you still didn’t get it, they’d explain it again.’’

These study groups encouraged just the sort of collaboration Francis had imagined. ‘‘It took the competition out of it,’’ Johnson said. ‘‘It wasn’t, ‘I’m mad because you got an A.’ It was, ‘How do we both do that on the next test?’ We had this feeling if we all stuck together and helped each other, we would make it.’’

Read the whole article here.