Yesterday and today on my commute, I listened to a GitMinute podcast focusing on Gerrit. As you might know, I am not a fan of Gerrit, so I was curious to hear what a big fan of Gerrit would have to say.
I was surprised. Luca, the show's guest, said that the Gerrit user experience was pretty bad, and that it was hard to use. Although of course he is a fan of Gerrit, he said that it is not appropriate for every team (especially small teams), and that it has a steep learning curve particularly for developers who are new to Git.
I was surprised that Gerrit's horrid email notifications weren't mentioned at all.
Back to work. On my way in, I stopped by the post office to mail a bunch of OVN stickers around the country and the world. People really seem to like getting them in the mail.
I managed to tie up a lot of loose ends in the morning, which leaves just the difficult parts for the rest of the week. (Is that a good tradeoff?)
I spent some of the afternoon working on a prototype of P4 for packet parsing in Open vSwitch. I'm presenting this at the P4 workshop next week, so I need something to talk about.
We spent the morning driving back home.
I managed to catch up on email a bit afterward.
Kate and Grace took some of the kids horseback riding, while Roberto and I took the others to the park in the morning. Ate leftover fajitas for lunch.
Went for a hike in the morning. Made fajitas for dinner.
Roberto and I went out in the evening. The original purpose was just to go to the casinos in Tahoe, but when I walked over to the Harvey's box office I saw that there was a concert followed by a comedy show, so I got tickets for both. The concert, by a band called “The Relationship,” attracted my eye because it had a member of Weezer. The comedy headliner was Tommy Savitt. I went in not expecting too much, because the last show we went to at the Harvey's Improv wasn't too good, and I was pleasantly surprised.
We drove to Tahoe in the afternoon, arriving in the evening, and met our friends at a rented cabin.
Today I attended:
“OpenDaylight and OpenStack.” This talk mainly showed all the details of switching network stack control from the OpenStack plugin to OVS over to OpenDaylight. In my opinion, it went too much into these mostly uninteresting details, so I didn't learn too much of substance.
“Taking Security Groups to Ludicrous Speed with Open vSwitch,” presented by Miguel Angel Ajo, Thomas Graf, Ivar Lazzaro, and Justin Pettit. The main point of this presentation was that the features coming in Open vSwitch 2.4 will allow security groups to be implemented in a much faster way than is currently possible with iptables.
“All Together Now: OSS Cross-Collaboration for a Network-Enabled Cloud,” a panel including Justin and some others. I found more interesting and informative than the OPNFV panel from the previous day, although I did ask a question at one point just to change the subject from a discussion between two of the people who had been on that panel.
I learned from this panel that OpenDaylight is not primarily an OpenFlow controller these days. Instead, it concentrates more on the management layer. Thus, there is a more obvious relationship between it and OVN than I had thought before.
Ate lunch with Justin and Kyle Mestery.
Flew home in the afternoon.
Today I attended:
“LXD vs KVM” presented by Ubuntu. I went into this thinking that they meant LXC, which is the code that Docker and other Linux-based container systems use. It was actually about a new system, LXD, that Ubuntu is building. LXD is a container system for “full machine” containers, that is, ones that run a full stack from “init” up, as opposed to the process containers that Docker uses. Yet another point in the design space.
“Neutron: VLAN Aware VMs,” a developer discussion. I was really surprised to see such a big crowd for this, probably over 100 people. It seems that there are lots of operators who want logical VLANs for their VMs; about half of the sizable number of operators in the room raised their hands when polled on whether they needed this feature.
It would not be too hard to add this feature to OVN. The main question is how to handle gatewaying to physical networks, since we're planning to subdivide gateway ports by VLAN. QinQ?
The main use case for logical VLANs seems to be NFV.
An OPNFV panel discussion. Dreadfully boring. I should have left after the first 5 minutes.
An informal meeting with Alessandro Pilotti about Open vSwitch on Hyper-V. He wants to make sure that OVN is built from the beginning to work well on Windows.
“Highly Available, Performant VXLAN Service Node.” I don't have anything to say about this, it was just what it said on the label.
“Docker Networking in OpenStack: What you need to know now.” This talk showed how to make VMs and Docker containers talk to each other on unified logical networks using a (presumably proprietary) PlumGrid plugin to Neutron. (OVN can do this too, and I didn't realize that it was special.) I kept expecting there to be more but really that seemed to be all of it.
Ate a fantastic dinner at Yew, in the Four Seasons hotel, with Justin Pettit and Bruce Davie and Thomas Graf.
Ate breakfast at Tim Horton's, with Alex Yip. I had not realized before that this chain is just like McDonald's.
I spent some time in the morning with Justin Pettit, Russell Bryant, and Kyle Mestery, watching them do a practice talk and giving a few suggestions. I had an idea from the slides (over 30 of them) that it would be difficult to fit it in a 40-minute slot, but in fact the run-through took less than 30 minutes, including a demo.
At 2:00 the actual run of the talk went well. I stood outside the door at the end and handed out OVN stickers. Lots of people asked for multiple to give out to colleagues.
Some Debian developers had organized a meet-up following a 3:40-4:20 session, but there was some kind of mixup and only two of us showed up. I met Clint Byrum, at least. He wrote down his GPG key signature and I agreed to sign it.
I went to a talk “Super Duper OpenStack and Docker Deployments” at 4:40 but it turned out to be an advertisement for a startup's monitoring software.
The OpenDaylight project hosted a very nice party at a restaurant named Al Porto. Justin and I attended and talked to several of the developers there. We wanted to smooth down some feathers, if any were ruffled—after all, OVN is a controller too and could be perceived as competing with OpenDaylight—but in fact everyone seemed to understand the difference.
Justin and I bailed early from the OpenDaylight party to attend a dinner among the OVN developers at a Japanese restaurant near the Hyatt. I was surprised to see Bruce Davie and Chris Wright there since I hadn't noticed them earlier at the conference (it turned out that they'd been delayed in travel).
This is the first OpenStack summit that I've attended. I am looking forward to figuring out how this community works.
I spent an unreasonable amount of time in the morning buying a phone SIM and trying to get it to work. Why do wireless carriers have the worst websites? The one from the 7-Eleven “Speakout” service actively fights against phone browsers. 7-Eleven wouldn't take my corporate credit card, either, so I'll have some trouble expensing it.
This is a very large conference. I think I heard that there are about 4000 people here.
There is less food than I normally expect to see at conferences. No breakfast, and not much food at breaks.
“Virtual Networking in OpenStack: Neutron 101” by Kyle Mestery and another presenter. I'm only starting to work with OpenStack a little bit, so it was helpful to get an overview.
“Building Managed Private Cloud for Enterprises with OpenStack and MidoNet” because on the OVN team we don't know enough about Midokura's products.
“Optimizing Contributions with Globally Distributed Teams,” a panel discussion. I think listeners new to the topic would get some benefit from this, but I have been doing distributed development for too long to learn much.
“Here be dragons—Translating OpenStack.” I learned about the Transifex web service for translation. It sounds like the primary benefit from this service, as opposed to the common practice of passing around .po files, is that it avoids having to submit the new .po files to the maintainers by hand. I had only heard of this tool previously in an email thread where Richard Stallman told GNU maintainers not to use it.
“The Big Tent—a look at the new OpenStack projects governance.” The sort of talk that I would normally avoid, but I figured that I need to know more about how OpenStack functions.
“Lightning Talks, Sponsored by HP.” Most of these talks were actually by HP employees or about HP Helion, which I found disappointing.
The booth crawl from 6 to 7:30 was really crowded. There was plenty of beer, but hardly any food. I spent the time peering at nametags for names that I recognized, but I did not see any, other than for people I knew well already.
After the booth crawl, there was a Neutron meetup in a neighborhood about a kilometer away from the conference. This was the most rewarding part of the day. I spent the whole time talking to various Neutron contributors. I was pretty surprised to find out that everyone knew about OVN already and were generally in favor of it; some were really excited, in fact. Open vSwitch has a positive perception, even though the OVS plugin to Neutron does not. (The Neutron developers clearly distinguish these, even though the larger OpenStack community tends to conflate them.)
It appears to me that OpenStack has a higher proportion of female attendees than RADIO or NSDI.
Flew to Vancouver to attend the OpenStack Summit. Vancouver looks like a nice city, in the dark at least.
San Carlos Hometown Days. We went to the parade in the morning and ate lunch at the food booths, then in the afternoon we helped out at the 4-H booth.
Still having some trouble with the home wireless network even after living here for 2 1/2 years. This afternoon I did a little research on wireless APs and decided that the 100 mW transmit power in the APs I have might be partly responsible. I ordered a new unit with 800 mW power. With luck, I'll be able to replace both current APs with this single unit somewhere in the middle of the house, with better results.
George Varghese came to VMware today to present some of his recent work and some thoughts for the future. I think that we might have some interests in common, so perhaps we'll try to collaborate or at least to meet again to discuss further.
I had a couple of meetings today with people about OVS and OVN. One of them was very promising—I think we might get some very useful contributions to OVS on the OVSDB front later this year.
Last day of RADIO. The other two papers presented by my team were today. I was session chair in the final session of the conference. The sessions seemed to be running consistently a little behind schedule, so I intentionally kept my introductions very short. “Our next speaker in the session is Ethan. Here's Ethan!” My philosophy is that it's about the speakers and we need to give as much time as possible to them.
This is the longest day of RADIO since it starts early and ends late. Following the evening events, a few of us from the Open vSwitch team met up with my friend Mike Dalton at Bourbon and Branch, a fancy cocktails place near the RADIO hotel. I had not been there before. The drinks were good, but I am not sure it makes up for the very slow service. The others who had been there before thought it was slower than usual. I had been careful not to drink much before arriving there, to avoid a hangover later, but I could have had a bit more since we only managed two rounds in two hours. Keeping things moderate isn't all bad anyway.
First day of VMware's internal “RADIO” conference. I enjoy this conference but I'm involved with it enough that most of the papers aren't really a surprise.
This year I had a booth in the Expo to explain OVN. We gave out stickers, went through a spiel, and answered questions. Today only, we also gave out pieces of cake decorated with the OVN logo. It was trouble getting the cake. First I walked to the nearest grocery store, a Whole Foods, but their decorators only work in the morning. Also: “We can't do anything in black. Black has chemicals in it. We only work with natural food dyes.” Whatever.
Then I walked farther along to the nearest Safeway. Safeway's decorators had also gone home, but the store was happy to sell me a blank cake and a tube of (no doubt chemical laden) black decorating icing and other supplies. I carried everything back to my hotel room and put the OVN logo on the cake myself. It looked OK.
One member of my team presented his paper today.
Mother's Day. I had given Kate her gift yesterday. She did not like it, so I've now arranged to send it back and get something else instead. Cassie's present, however, was more successful: chocolate is always welcome!
Had Lucila and her family to dinner for the first time in forever. I made jambalaya, which Lucila says is a lot like the Brazilian dish moqueca. Strawberry-rhubarb crisp for dessert.
Finally got the car back from the shop, after two week. The dealership replaced the compressor, the evaporator, the condenser, and numerous tubes and hoses. I am so glad that this is warranty work.
The full OVN stack is now checked in, and today I checked in a fix for a bug that had made it in due to lack of testing for revised patches. I tested that it all works now.
Back from NSDI. Lots of meetings.
Last day of NSDI. I didn't have much patience with the talks.
Got delayed on my way home by BART delays due to a broken track near the Civic Center station.
My favorite talks of the conference were today: Succinct: Enabling Queries on Compressed Data and PCC: Re-architecting Congestion Control for Consistent High Performance.
Justin and I hosted an OVN BoF in the evening, but few people showed. We had informal conversations with those who did.
I accepted the “best paper” award for the paper “The Design and Implementation of Open vSwitch” on behalf of the 12 authors this morning at USENIX NSDI 2015. I delivered the talk just after lunch, at 2pm.
I spent more time working on my presentation for NSDI. Teemu gave me a lot of great feedback on my slides, that should make it a much better talk.
Kate is really unhappy that I'm going to be away so much this month.
I spent much of the day preparing for my NSDI presentation on Monday. The result seems like it will be much better than the hastily thrown-together version I used for a test run yesterday at work.
Kate and Cassie spent the morning at a scavenger hunt at Happy Hollow in San Jose. (They won! Best time out of 7 teams.)
In the evening Kate and I attended a murder mystery dinner party. I had the role of Banker Bob and Kate the role of Banker Bonnie. My role was a bit of a stretch for me. As my character sheet said, “As the bank owner of Deadwood, you are a businessman first. Your bank and your money come before everything, including your wife Bonnie.” I had a hard time ignoring her most of the night, especially since her character sheet said that she would do “anything” to get my attention. Both of us had a lot of fun.
OVN works! I tested it and fixed a couple of bugs that I found, but it was basically working based on the patches from last night. (Another developer had claimed that he got it working before my patches, but we later figured out that this involved a mistake.)
We received an update from the car dealership: the AC compressor exploded and sent metal shrapnel through every rubber belt or hose nearby. All of them need to be replaced. It will take until next week. We are surprised: this is a 2014 car with less than 10,000 miles on it.
I posted patches that complete a demo milestone of the end-to-end OVN system. Not yet tested.
I spent my morning at the Stanford Computer Science Department 50th anniversary celebration. I remember Dan Boneh's talk the best. I found his slide on the “crypto wars” insightful. Quoting:
First crypto war (1975–6): Attempt to classify open crypto research. Outcome: first amendment triumphs.
Second crypto war (1994–5): Attempt to enforce key escrow (Clipper chip). Outcome: public outcry triumphs.
Third crypto war (2015–6): Attempt to require backdoors in every encryption product. Outcome: ???
He also pointed out the sequence, predicting a fourth crypto war in 2035.
I went to the Stanford Computer Science Department 50th anniversary dinner in the evening. It was pleasant to speak to professors and former students I knew from when I was a student, and to others who were before (or after) my time.
When I left to bike home, the department chair (Alex Aiken) was getting on his bike too. The idea of riding a bike in a suit and jacket seemed unusual to me, and I said so: “Don't you ruin the pants?”
“No, I tie them up.” Well, yes.
“But what about the seat?”
“Oh, every couple of years I'll hear a tearing sound in class, and then it can be a little embarrassing for the rest of the day.”
I made a strawberry-rhubarb crisp in the morning. Delicious.
I took Cassie to her hockey practice. Afterward, the car stalled repeatedly pulling out of the parking lot. I pulled into the lot at the next business down the street and called the Ford assistance number for a tow. The tow truck driver took me to the nearest dealership, where I put the keys into a night drop box (being Sunday, they were closed) and took the train home.
The tow truck driver speculated that the AC compressor belt had failed, since when he turned off the AC the car seemed to behave better.
We spent the evening at a party at a neighbor's house. Nice people.
I posted patches that were much improved from the RFC ones from Wednesday.
I helped give feedback on presentations for VMware's internal “RADIO” conference for much of the day.
Worked from home again to keep up my momentum, but I was derailed significantly by reviewing financial documents.
In the evening a pair of houseguests we had been expecting arrived quite late, around 9:30. We greeted them and let them go to bed. They were gone (to the airport) at about 4 am the next morning, before any of us were up.
I posted OVN patches that come close to a complete working system in RFC form at the end of the day.
I worked from home and got a ton done on OVN.
Felt a bit better so I went in to work.
Feeling sick, maybe worse, today and did not want to do anything active. I did help slightly with the cleanup day at Cassie's school.
I finished building the shelves on my fancy cat tree.
I got the basic logical pipeline table generation code working for OVN. We're getting closer.
I was still feeling pretty sick, but I had meetings that I didn't think I should skip so I went in anyway.
Leslie Lamport came to VMware in the afternoon. After the main VMware gathering in the auditorium, there was a special conference with Leslie for just the principal engineers. This is my favorite PE perk.
I have not been feeling entirely well for a few days. Today I was feeling really sick. I worked from home.
I got the OVN expression code checked in. I am really happy—this code is elegant and it should make other parts of OVN easier.
Back to the office. I tried to catch up on email.
We flew home.
The morning highlight today was a horseback ride for all of us, with Betsy and Jess, across the battlefield at Gettysburg. Our guides told us something of what they knew. It was not a full tour and we did not go far, but we enjoyed it.
We ate lunch at the Appalachian Brewing Company.
In the afternoon Betsy and Jess gave us a tour of Gettysburg College, where they work. It is full of beautiful brick buildings, many of which date from the Civil War.
We drove back toward DC in late afternoon and stopped at a hotel. Kate returned the car while I took Cassie to the pool.
Our day began with the Cherry Blossom Festival Parade. We took the metro to the parade. We had pre-purchased bleacher tickets. We watched most of the parade. We felt restless and wanted to avoid the big rush to leave, so we took off a little early and made our way back to the hotel. Ate lunch, then we picked up our rental car and drove to Gettysburg, which was a 2 or 3 hour drive.
We arrived at Betsy and Jess's apartment in late afternoon. We spent a long time playing charades based on a card deck that Betsy gave Cassie for her birthday.
We all ate dinner in downtown Gettyburg at a Chinese restaurant named Ping's.
The three of us went to our hotel. I took Cassie to the pool before we all went to bed.
We visited the National Zoo in the morning. After lunch at the zoo, we made our way to the National Mall for a tour of the monuments. We visited the Lincoln Memorial and the Korea memorial. I don't think I had been to the latter before, or at least I did not remember it. It was moving, with the statuary of the soldiers walking through the rice paddy.
We walked through the FDR memorial park and around the Tidal Basin to the Thomas Jefferson memorial, then to the booths at the Cherry Blossom Festival.
Everyone was tired, so we went back to the hotel to rest for the afternoon.
In the evening we met Olivia at the Hard Times Cafe in Alexandria. (The DC metro serves northern Virginia as well as DC!) After dinner we went to a custard shop for dessert, which was altogether too much food. Kate and Cassie went back to the hotel on the metro. Olivia and I stayed out for a while to have a drink at a local bar, then I went back to the hotel too.
We visited the Library of Congress, Jefferson Building, in the morning. The exhibit showing Jefferson's original library made the greatest impression on me. It's not very big by modern standards but it's diverse and it gives the feeling of history.
After that we proceeded to the Supreme Court. I was surprised to see a depiction of Muhammad in one of the friezes. We had an early lunch at the Supreme Court cafeteria. It was a real find, some of the best food on our trip, lots of home-cooked food. I had a homemade cabbage roll and roasted yams, and Kate and Cassie shared slices of turkey cut for us right off the bird. All for about $6 a person.
From there we went to our congresswoman Jackie Speier's office for a tour. I was really impressed with the friendliness of her office staff. I know that they greet hundreds of constituents a month but they still act like each visitor is someone special. All of her staff took time to tell us about themselves and what they do. After bagels and coffee one of her interns took us through the underground tunnels to the Capitol, where we were privileged to skip the long lines for tours. After the general Capitol guided tour, we used the passes that her office had provided to visit the House and Senate chambers.
We met Olivia in mid-afternoon. We had cupcakes at a cafe to celebrate Cassie's birthday. Afterward, we toured the botanical gardens with Olivia, then ate dinner together at a Thai restaurant. Cassie lost a tooth during dinner. She was very proud.
We ate breakfast at Dunkin Donuts near the hotel. I don't eat much fast food, so it was actually somewhat novel.
We spent the day at the Smithsonian. We started out at the Smithsonian “Castle” that acts as a sort of headquarters for the museums. From there we went to the Air and Space Museum, then after lunch to the Natural History Museum. I was most impressed with the original Wright Brothers airplane at the Air and Space Museum, even though I had seen replicas elsewhere. The Wright Brothers are some of my greatest science and engineering heroes.
Had dinner at an Ethiopian restaurant named Meskerem with my friend Olivia, who lives close to DC. Afterward Olivia and I went to the hotel bar for a couple of drinks, then Kate and I traded off and Cassie and I went to bed.
Flew to Washington DC with Kate and Cassie.
Scrambled to tie up loose ends before going on vacation with my family.
Any number of people have contacted me about my thoughts for the ONF Extensibility Working Group and its future, now that Jean Tourrilhes has resigned as chair and left ONF. Here are my current thoughts, to give me a place to refer people in the future.
First, I don't think that anyone is going to be able to step into Jean's shoes. Jean was so thorough, so knowledgeable of the OpenFlow specification and its history, and so willing and able to use a great deal of his valuable time for the working group, that no one can match it. Simply on that basis, the extensibility working group is going to change.
Second, we now have numerous ways to extend the spec without changing the main specification. This means that working groups outside extensibility can do their work without needing to go through extensibility. There is this perception that extensions are second-class citizens, compared to what actually goes into the main specification. I think that we need to fight against that: extensions, when released through a reasonable process by a reputable group, are just as legitimate as incorporated pieces of the main specification.
Finally, I think that it's OK to slow down. It would not be the first time that we've had a period during which only bug fixes go into the spec. The new features being proposed now are pretty limited, and vendors need time to catch up to 1.4 and 1.5 anyhow.
Major progress on my fancy cat tree. I wove and stapled vinyl into a pattern along the post, punching holes to expose the holes in the post. I built and tried out a couple of prototype cat shelves, finally just before bedtime coming up with one that worked out OK. The cats were happy with even the prototypes that I rejected, though.
Today was Jeremy Stribling's last day at Nicira/VMware. He was one of our early hires, and one of our distributed systems experts. We had a farewell lunch in his honor at Steins in Mountain View. End of an era II. Too many of those coming along lately.
Jean really has left ONF now. There was no Extensibility WG meeting this morning because OpenFlow 1.3.5 was released, which was his final goal, and thus he very quietly stopped organizing the meetings. End of an era.
In the afternoon Justin and Guru and I went up to San Francisco for a meeting with Madhu from Socketplane (now acquired by Docker). Docker has beautiful offices up there, I'm jealous (and I say that as working from the rather nice VMware offices in Palo Alto). We had about a 90-minute meeting to talk about integrating OVN with Docker. It sounds like there are two important steps. First, for basic integration, Docker has to implement its proposed network plugin architecture, which is planned for Docker 1.7. Second, for better integration, Docker needs to support labels on its network endpoint objects (though we can work around this until then).
Worked from home all day and managed to send out some patches for converting Boolean expressions to OpenFlow flows in OVN. Pretty happy with them.
I made the ginger pork with gojiberries recipe for dinner. It's a house favorite around here ever since I received a clay pot for Christmas a couple of years ago. I don't follow the recipe exactly: I substitute cranberries for gojiberries and any cooking wine I have on hand for the plum wine. It's delicious. I wonder whether it would be better with the “real” ingredients, but it's hard to imagine how.
I put the foam on top of 3/4-inch plywood and upholstered it, around the hole, using staples to hold the fabric in place. No problem. I did have to cut half an inch off two sides of the foam because my plywood was only 23 inches wide. Unfortunately I poorly chose a marker for marking the place to cut so now I have some blue marks bleeding through in few places on the fabric. Oh well, the cats won't care.
Spent a lot of the day at home working on my fancy cat tree. I drilled 17 1-inch holes through a 4x4, plus 17 half-inch holes perpendicular to them, generating a huge pile of sawdust in the garage.
In the evening Kate had her hockey championship game. Her team won the “Stanley Keg.” Cassie and I went to watch.
I dropped by House of Foam in Palo Alto and bought a 2-foot square chunk of 3-inch extra firm upholstery foam for the base of the cat tree. Good upholstery foam is expensive (about $75 after tax for those 4 square feet) but it's comfy as heck. There's no way you can bottom out on 3-inch foam even if you just rest one knee on it. Bob Tallman at the store kindly cut a 3.5x3.5 inch hole in the center of the foam for the cat tree post.
Second day of the promotion reviews. Only a half-day. In the afternoon I managed to get some work done.
In the evening I got an inspiration to build a fancy cat tree over the weekend, so I went out to Home Depot and Jo Ann Fabric and bought a big pile of wood and fabric and paint. Let's see how it goes.
Spent all day in a meeting with all the other VMware principal engineers, going through candidates for promotion to senior staff engineer and principal engineer. It was less stressful and boring than I expected. Everyone up for promotion is very good, and we approved most of the promotions.
As a matter of habit, I always come to a complete stop at stop signs (on my bike). Since no one else does, I always get to go first. Sometimes this makes people mad. I think that a lot of drivers think that drifting forward slowly is the same as stopping.
Continuing to work madly on OVN Boolean expressions, making excellent progress on writing tests. I'm taking an exhaustive testing approach, by writing code to generate all possible Boolean expressions (up to particular limits) and testing that the various necessary transformations on them do not change their values. I've found and fixed tons of bugs, including many memory leaks, this way.
We put the rest of the plants into the ground in the afternoon. It's much easier digging in the yard than last summer, because we've added 6 inches of soft compost and wood chips. Really quite easy now. The soil is amazingly moist when you dig down an inch or two, despite our long spell of dry weather.
In late afternoon Cassie and I biked to Burton Park and played bocce.
A shipment of plants arrived yesterday, so we put them plants into the ground, then in the afternoon Kate went off to a nursery in Half Moon Bay to get some more.
Wow, all three of the papers that my group submitted to VMware's internal RADIO conference were accepted as short or long talks. I can't say how impressed I am!
I took the afternoon off from work and we all went into San Francisco for a last day together. As late afternoon approached, we took Lejna to the airport, ate dinner, and saw her off. Goodbye Lejna!
Starting to get really annoyed the number of on- and off-list email inquiring about documentation and examples on how to integrate Open vSwitch with hardware ASICs. These inquiries usually come from employees of ASIC vendors. There is no documentation and no examples for these purposes specifically because the policies of these ASIC vendors prevent them from being published! You guys fix your policies, and then people will start publishing code for your chips. Don't complain to me: it's your fault and you, not I, have the power to fix it.
Lejna's departure is coming up in two days. Everyone is starting to mope and moan a bit about it. We'll miss her.
I took a Schlage lock cylinder that I had mis-keyed into work with me today. I took it to the lockpicking interest group and, after about 30 seconds of instructions, spent about 5 minutes fiddling with a tension wrench and a pick and sprang it open. I carefully popped off the retainer ring and fitted a rekeying cylinder into it.
Locks really aren't that secure if a rank amateur can open one in 5 minutes with a couple of bent pieces of metal.
We hosted a going-away party for Lejna at our house. A lot of her friends that none of us had seen in a long time came to the party. We had a lot of good food, but the best was a strawberry-rhubarb pie that Kate made. Probably the best I've ever had.
Mailed my tax paperwork to my tax preparer. Not too happy with them because they say they're charging an extra $75 just to consider the new ACA healthcare laws; that should take them about 3 seconds with me since I have employer-provided group coverage. They also charged me an extra $240 last year for estimated taxes. If they try to charge me for both of those this year, then I'm going to find someone new.
Pi Day Eve (and Friday the 13th). VMware gave away “Pi”nt glasses at their Friday afternoon celebration. I stopped by just long enough to pick one up.
Got a prototype implementation of code to convert Boolean expressions into OpenFlow flows working in OVN. I'm happy; it seemed like it was taking too long. Still a lot of work to get it finished and bug-free though.
Program committee meeting today for VMware's internal RADIO conference. All three of the papers that I coauthored were discussed, although I don't know the decisions. A very long day: the discussions ran from 8 am until about 6:30 pm.
Sunday night, I introduced Lejna to the Rocky Horror Picture Show. Back when I was a teenager I learned all of the audience participation lines, and so at the beginning, at least, I was calling out pretty often, but I quickly got tired of doing it alone.
Kate and Lejna think Rocky Horror is pretty weird; they're not wrong.
Got familiar yesterday with what might be a new chore every year or two. About a week ago, we noticed that the swinging glass door on the shower in the master bedroom had started to hit the fixed frame. Not good—if it hit too hard it would surely chip or break. Yesterday I finally looked up the procedure for adjusting it. Loosen the screws top and bottom (required an unusually large Philips #3 screwdriver), adjust the position carefully, tighten again with screwdriver. The door is heavy and unwieldy enough that I had Kate do the tightening while I held it in place.
I responded to a couple of surveys from ACM on open access to research. I'm so glad to see that ACM is thinking about opening up its publications to the world. It's a “moral imperative” in my opinion, in the words of an answer to one of the survey questions. USENIX manages to do this, without charging those who submit papers, without charging ridiculous membership dues, and still retains high quality conferences and submissions, so I don't know why ACM can't do the same thing.
This evening was the annual Spring Gala from the San Carlos Education Foundation, which supports the San Carlos public schools. Kate and I won the auction for a vacation for 8 in Montana at a dude ranch. The trip has to be between November and April, so I guess we're in for a winter vacation.
Went out tonight with Mike Dalton, a friend from Stanford, to Dave & Busters in Milpitas. We ran out of games we wanted to play pretty quickly—they have less variety than I remembered from years ago. It's pretty much all shooters and racing games, other than the ticket games that are really only fun for kids. I do like shooters and racing games, but I can only play them for so long. The best game was a new Star Wars game with a pretty immersive view, although the controls didn't seem too responsive.
Getting a lot done on OVN. Hope to get out another patch series next week.
Jarno and I biked home together, as we do from time to time. Jarno has a brand new carbon fiber Cannondale with fancy electronic shifting. It's noticeably lighter than my aluminum Trek from 2002. I think we were about the same speed before (when he was on a $300 hybrid), but now I have to work to keep up. It has a built-in power meter. He reported that we were (well, he was) using 300 W to go up the first hill and that he got up to 450 W briefly on another. I'm surprised—I had an idea that only pros could put out over 400 W.
Back to code reviews today. I've been ignoring them for a while to get some coding done on OVN, but they really piled up. I spent the day on reviews. The most significant result was that I was able to commit the SPBM Auto-Attach support from Dennis Flynn at Avaya.
I spent a lot of time over the weekend thinking about how to reduce the general Boolean expressions that I want to use for flow matches in OVN to the form that Open vSwitch supports. I think that this comes down to what one might think of as a nonstandard “normal form” that so far I think of as “disjunctions of conjunctions of disjunctions.” Today, I had a good conversation with Tim Hinrichs, the best guy I know in logic, about the possibilities. I was a little surprised to learn that this isn't something that I can just read a bunch of papers about. Maybe, if I come up with a good solution, there'll be a paper to write about it, who knows.
Spent most of the day with some friends, first at Coyote Point and then later at home. We rarely see these friends but when we do it seems like we always spend a long time together.
Finally got a little way into The Art of Community only to realize that all of the figures are missing. A web search shows a few other people with the same problem. Fortunately none of them so far seem important to the message but only really for amusement.
Had a meeting at Mesosphere. They helped us to understand how the container model works in the environments that they work in. It also sounds like OVN is a pretty good match for what they need for DCOS. I was pleased to see Christos Kozyrakis again; I don't think I'd run into him since I was a grad student years ago.
It's so great going to go meet people at startups. The energy level and the excitement is so high, I really enjoy it. And Mesosphere is in a nice building, only a short walk from Caltrain. Some day…
Went to listen to a Customer Advisory Council session on NSX this morning. When I saw one of the attendees, it reminded me of how the classic example of separation between tenants is always that Coke and Pepsi don't want their opposite number to be able to spy on their secrets—one of them was in attendance!
Kathy Griffin came to VMware today for an hour-long interview. Kate came to campus for it and we sat in the second row. I like a lot of Kathy Griffin's humor, although some of it goes over my head because I don't keep up with celebrity news. Before the interview I had not realized that she had been on TV (“Suddenly Susan”) or that she was such a trailblazer among female comedians.
Kate and I went out to dinner afterward at Parkside Grille in Portola Valley. I was surprised to find a non-lumber-based business so far off the beaten path.
I've spent the last couple of days of actual coding time (as opposed to work time) working on a lexer for OVN, to use as the bottom layer of the parser for matches and actions. The strategy that OVS uses for parsing OpenFlow from strings is terrible. I don't want to repeat that mistake with OVN. Maybe it'll even be possible, eventually, to use some of the same code to parse OpenFlow.
The chicks are now 5 days old. They have grown visibly since we first got them at 1 day old.
Since Lejna has to leave for Bosnia again soon, we want to spend her last few weeks doing as many fun things as we can together as a family, so in late afternoon we went to Dave & Busters in San Jose. We played games for an hour or too, then had some dinner and went home.
I shingled the roof of the chicken coop, with the cedar shingles that I bought last weekend. It took a few hours because I had never done anything similar before, but it wasn't really difficult.
I spent much of the day making up a fairly detailed task list for OVN, and then I published it, getting some useful feedback already.
I got the OVN design documents to the point where I was comfortable publishing them for public discussion. I received more comments than I expected, and quite quickly.
In the afternoon I attend “Unconscious Bias Training” in VMware's cafeteria with a lot of other people. I was nervous about whether it would be useful or a waste of time, but I was pleasantly surprised. Given what I was focusing on, it was a surprise to get a text from Kate in the middle saying “Going to get chicks now.” Indeed, when I got home, there were a couple of chicks under a heat lamp in a plastic bin inside the chicken coop.
I had a really productive day getting together some OVN design documents for internal review.
Kate and I went out for a date to John Bentley's, a nice restaurant along El Camino in Redwood City.
Kate's parents left for the airport around noon. I spent late afternoon first buying lumber and shingles for the chicken coop, then shopping around town for a trunk or storage bench with a lock on it. I had pretty much given up on finding what I wanted until I stopped at an antique store (“Antiques Then & Now”) near our house. They had a lovely old cedar chest with built-in felt-lined compartments and a lock (with the key!). The price was right so I took it home.
Kate and Cassie and Grandma Pam went out for much of the day. Dave and I spent an hour or two mixing and pouring cement into some gaps between boards by the chicken coop. It will provide a step up for Cassie and keep chicken poop out of difficult-to-clean spaces.
Valentine's Day and Kate's birthday. Birthday presents (those diamond earrings I bought last week!), chocolate, lunch out. I made Irish stew for dinner (Kate's request). I took Kate's parents to watch her hockey game at 10:30 pm.
Checked in a fix for a bug that was reported via the Open vSwitch security mailing list as a vulnerability. (The commit is titled “mac-learning: Implement per-port MAC learning fairness.”) I am not convinced that the bug itself is that big of a security vulnerability, because it is only a problem if users regard MAC learning as a security feature rather than an optimization. I spent a bunch of late-night time on fixing it on Wednesday night, though, because it seemed to me that it was a serious publicity vulnerability: the bug's “security researcher” reporter seemed a bit breathless about it, and had already given a talk about it at a security conference back in September (!). Also, I had a query from CERT about the whole thing. Since it was in my power to fix it, without too much trouble, that seemed like the best solution.
The solution adds a lot of mildly tricky code. A bug could be bad. The fix was reviewed by a few very good programmers, though, so I think that the risk is minimal.
I think that the fix is better than what typical physical switches implement. I like that it is adaptive, that is, there is no need to have any settings or knobs for users to tweak, it “just works” out of the box.
I wish that the reporter had let us know before making this public, instead of months afterward. (Would I have responded as quickly? I don't know.)
Kate's parents arrived today, to stay for the long weekend.
Our new(er) car has heated seats. I didn't think they'd be useful in our area with such mild weather, but they were part of some package that included other features that we did want. In the last week I've found that they're nice sometimes though. One day I was a bit wet from rain and the heat (at the lowest setting) felt nice and helped to dry my pants. Today, my back was a little sore, and the heat on my lower back was very soothing.
Kate showed me in the morning that yesterday she had attached the front and roof to the chicken coop. Now it just needs some finishing touches: chicken wire over ventilation holes, shingling the roof, and I don't know what else.
The loud weird sounds we heard last night were revealed today to be a pair of raccoons mating outside our bedroom window, on top of the garage. I hope they don't mess with the wind vane.
Went to an event organized by Amplify Partners in the evening, at Gravity in Palo Alto. I've been thinking a lot lately about the lack of women in (my part of) the tech industry. This kind of event really exemplifies it: I did not notice any women in attendance at all. I did enjoy it, though.
Lunch with Yatish Kumar from Corsa. Hi Yatish!
The rain continued. Kate and I spent most of the afternoon working on the chicken coop. We got the floor and the sides built and attached, and the front ready to go on. At that point it was dark and still raining, so we called it a night.
Two very different kinds of shopping today. In the morning, Lejna and I went to Shane Co to buy diamond earrings for Kate's birthday (after all, she had them pierced just for me). In the afternoon, all of us went to Home Depot to buy plywood and other supplies for building a chicken coop.
Hard rain most of the day. In the morning I helped set up the school book fair, then due to the rain I drove into work. Finished up the NSDI revisions and sent the paper in to our shepherd. Spent some time discussing the OVN database schema with Justin. I left the office about 5:10 and due to traffic it took about an hour to get home, about twice (three times?) as long as going the opposite way at 10:30.
A long day. I took Caltrain to meet Dan Talayco in Mountain View for breakfast, where we talked over our ideas about ONF and P4. Then I took Caltrain back north to Palo Alto, where I assumed I could catch the next Marguerite or VMware shuttle to campus. I was wrong. It was just before 10 am when I arrived, and all the shuttles and Marguerites had already gone for the day. I gave up on working from my office and just went home to work from there. (I'm not a big fan of working from home because then I end up working all day long instead of just when I'm in the office.)
Today I realized that revisions for the team's NSDI paper are due to our shepherd on Friday, and that I essentially hadn't done anything yet. I scrambled toward getting it done.
Kate and I visited our new financial advisor for an initial meeting. It's easier than I expected, so far.
Came back home from the leadership summit.
Passing through security at the New Orleans airport and preparing to collect my things from the conveyor, I had to basically dive to grab my laptop and my phone, which were about to hit the floor because the bins they were in were flipping up and over due to incompetent an X-ray operator. I grabbed all my stuff and walked on, and then 100 yards away I realized that I was missing my watch. I walked back but it was gone and the TSA personnel were less than helpful. Assholes.
Kate and Lejna picked me up at the airport. Kate had had her ears pierced! I asked her a while ago to do it. I like the way it looks.
Final day of the leadership summit. More speeches and Q&A sessions. We finished up the project planning exercise with Afterburner. A Superbowl Party was held in late afternoon.
The summit took up all day, from 7:30 in the morning until 10 at night. The breakout sessions on planning a “mission” in teams, by Afterburner, were much more entertaining and also more useful than I would have guessed. I especially appreciated their discussion of the value of their debrief advice: always do a debrief, whether the mission is successful or not, and keep it nameless and rankless so that you get honest feedback.
Today's highlight was the evening event. At 6:30, a marching band came parading along Canal Street, and all 200 of us at the conference joined in behind it. We followed the parade along Canal, turned onto Bourbon, then a few more turns and finally onto Royal where the band directed us into Latrobe's, where we had a fantastic 3-course dinner.
I spent the day up to 1 pm working on finishing touches for two SIGCOMM papers (the deadline was 2 pm). By then I was starving, having skipped breakfast, so I walked down to the river along Canal Street and walked along the boardwalk into the riverside mall, where I ate crayfish étouffée at a local (seeming?) place in the mall food court.
The conference got under way at 5 pm with a reception followed by speeches that lasted until 9:30 pm.
Traveled to New Orleans for VMware's internal leadership conference. Got in late after some delays in the airports on both ends. Ate dinner with a couple of colleagues. Feeling a bit rundown from a cold, so I went to bed without exploring the nightlife.
Tonight at midnight was the deadline for research paper submissions to VMware's internal RADIO conference. My team submitted three papers. I had my doubts about some of them for a while, but I think that all three of them turned out strong. If it works out, maybe some of them will turn into public papers in a while.
Recording audio wasn't much harder than writing text, and I tended to do it in the same place and at roughly the same time that I'd write text anyway. But I had an idea that I'd start getting new inspirations or make different use of an audio medium than I would a text medium. (Also, I had just bought a USB headset anyway for to be on the Packet Pushers podcast.) It didn't happen. So maybe I'll give it another shot later, but I think I'm done with audio for now.
About the OVN announcement.
Some rambling thoughts that bubbled up today.
Packet Pushers Podcast about Open vSwitch.
Books of 2014:
John Scalzi, “Red Shirts”, 2012, 978-0-7653-1699-8, 5B.
Sinclair Lewis, “Babbitt”, 1922, 0-451-51897-7, 9B.
Dick Francis, “Banker”, 1982, 8, 0-449-20262-3, 8B.
Ivo Andric, “The Bridge on the Drina”, 1945, 0-226-02045-2, 8B.
John le Carré, “The Spy Who Came In From the Cold”, 1963, 0-7434-4253-9, 7B.
John Evangelist Walsh, “One Day at Kitty Hawk”, 1975, 0-690-00103-7, 9B.
Ben MacIntyre, “Operation Mincemeat”, 2010, 978-0-307-45328-0, 7B.
Amanda Cross, “In the Last Analysis”, 1964, 0-380-54510-1, 4B/3D.
John Scalzi, “Fuzzy Nation”, 2011, 978-0-7653-6703-7, 9B.
Umberto Eco, “Foucault's Pendulum”, 1988, 0-345-36875-4, 6B.
Lawrence Block, “Sometimes They Bite”, 1964, 0-515-07407-1, 7B.
James White, “Tomorrow is Too Far”, 1971, 5B.
Dick Francis, “Proof”, 1985, 0-449-20754-4, 8B.
Kirk Mitchell, “Cry Dance”, 1999, 0-553-57914-2, 6B/6D.
Dick Francis, “Come to Grief”, 1995, 0-515-11952-0, 7B/8D.
Poul Anderson, “Orbit Unlimited”, 1961, 6B/7.5D.
Sue Grafton, “R is for Ricochet”, 2004, 0-425-20386-7, 7BD.
Dick Francis, “Come to Grief”, 1995, 0-515-11952-0, 7B/8D.
Sue Grafton, “U is for Undertow”, 2009, 978-0-425-23811-0, 6B/4D.
William Poundstone, “Prisoner's Dilemma”, 1992, 978-0-385-41580-4, 7B.
Janet Evanovich, “Visions of Sugar Plums”, 2002, 0-44903-00699-8, 5BD..
Tim Dorsey, “Hammerhead Ranch Motel”, 2000, 0-380-73234-3, 5B/5.5D.
Carl Hiaasen, “Skin Tight”, 1989, 0-449-21941-0, 7BD.
George Eliot, “Silas Marner”, 1861, 7BD.
Ron Hansen, “The Assassination of Jesse James by the Coward Robert Ford”, 1983, 8B.
John Irving, “The World According to Garp”, 1976, 0-345-41801-8, 7B.
Lilian Jackson Braun, “The Cat Who Moved A Mountain”, 1992, 0-515-10950-9, 6B.
Peter Carey, “True History of the Kelly Gang”, 2000, 0-375-72467-2, 8B.
John D. MacDonald, “One More Sunday”, 1984, 0-449-20703-X, 6B.
Harry Kemelman, “One Fine Day the Rabbi Bought a Cross”, 1987, 0-449-20687-4, 6B.
Lilian Jackson Braun, “The Cat Who Saw Red”, 1986, 0-515-09016-6, 6B/6.5D.
H. G. Wells, “Collector's Book of Science Fiction”, 1987, 978-0-89008-208-8, 8B.
Dick Francis, “In The Frame”, 1976, 0-449-22116-4, 7B/6D.
Lawrence Block, “Time to Murder and Create”, 1976, 0-380-76365-6, 5B.
Paul Levine, “Fool Me Twice”, 1996, 0-380-72590-8, 5B/7.5D.
Nicholas Meyer, “The Seven-Per-Cent Solution”, 1974, 345-24550-4-195, 6B/8D.
Neil Gaiman, “The Graveyard Book”, 2008, 978-0-06-053094-5, 6B.
Dick Francis, “Enquiry”, 0-671-44926-5, 6B/6.5D.
Noel Malcolm, “Bosnia: A Short History”, 1994, 978-0-330-41244-5.
James White, “The Dream Millennium”, 1974, 356-24012-X-125, 5B.
George Orwell, “Down and Out in Paris and London”, 1933, 8B.
E. E. “Doc” Smith, “The Skylark of Space”, 1928, 7B.
2014. 2013. 2012.