LinuxWorld Expo Report, Part 3
~ by Marty Connor

7am comes quickly when you stay up until 3am the night before.

I didn't know exactly where my strength would come from that first morning of LinuxWorld Expo, but it was now Tuesday morning, when the show would open, and I was determined to give it all I had.

I was surprised at how good I felt, actually. Michael and I had made good progress on our debugging the night before, and though we didn't get as far as we would have liked, I think that the process of debugging together was as important as the result. You get to know people when you're both tired, under stress and time constraints. You learn what is important to each of you, and you learn to trust.

I am sure that many who read this are no strangers to late nights. Whether you are staying up late working on a program, a case, a presentation, or a budget, you must face the morning, and the divine keeper of sleep will eventually claim what is owed. One may delay payment with stimulants, internally or externally supplied, but the loan of time will be repaid, and with interest.

7am. I wake, still in my clothes. Eyes dry, but mind alert. It's really not as bad as I thought. I don't generally do caffeine; it's too strong and uneven a stimulant for me. Instead I try to stay hydrated, and electrolytically balanced. I figure if the task is fascinating enough to keep my interest, I'll find my way.

Anselm got in late the night before, and will be here soon. Michael took a cab back to his hotel, and I don't know his state, but I do know that we need to get around 200 pounds of gear to our booth, and set up by 10am, have it up and running, and be ready to demo.

Shower time. In my head I'm doing calculations. Show floor opens to exhibitors at 8am, that gives us two hours to get set up once we hit the floor. Maybe a quick breakfast? Probably not. I have some health-food-gatorade-like-beverage and a bag of granola, and eating the wrong things (or too much) will only cost energy for digestion. I like to be a little hungry when performing -- it helps me keep focus.

7:20am. A knock on the door. Anselm is here. Wonderful. I quickly greet him, and finish dressing in the bathroom as I explain the situation. He's been to LinuxWorld Expo before, so he understands the basic concept. We need to repack the shipping case. The rackmount system needs to come with us. Cables and peripherals are strewn everywhere.

I decide that to save time it's best to just put as much as possible in my rolling suitcase, and we'll sort it out later on the show floor. Cables we understand and can easily disambiguate.

8:05am. Michael calls on the cell. He's running a bit late and suggests we meet at the Moscone convention center. Since we're still packing, I tell him we'll wait for him since we can use the extra hand wheeling stuff the block and a half to the show.

8:45am. Michael arrives. We're pretty much packed up, and after quick introductions we roll, literally.

This is LinuxWorld 12, but I still carry some of the excitement of the first one. This is a worthy hack, bringing everything together in this way. People, hardware, software, signage, flyers, and the confident feeling that I am with people who not only know what to do, but who have come this far, and paid this much, by their choice, to share this experience. I live for this.

9:00am. We reach the Moscone. Time for a small efficiency hack. Anselm doesn't have a badge yet. Michael and I do. I have to make the badges. So, for a few moments, Anselm becomes me, and he and Michael take the gear to the booth. I stand in line, create Anselm's badge, and become Anselm for a few moments to get to the booth. Every moment counts. Our little concurrency hack works.

9:10am. I reach the booth. Michael and Anselm have made splendid progress. The booth is starting to look like a booth. We say hello to our neighbors, Creative Commons and Debian, as well as the LTSP.org gang. Sometimes I bring coffee and doughnuts for the early birds, but no time this morning, so hearty welcomes are all we can offer, and they are sufficient.

9:30am. Banner time. Our banner is a 3x8-foot affair, laminated. I like to put it up the day before because it has curl, and takes a little while to relax. We hang and clip it, and it's a little curly around the sides, but it will do. I remind myself (as I seem to recall doing for the last two shows) that next show we'll get a fabric or vinyl banner that folds and packs more easily. Instead, we used the money to get a USB drive this time. We'll manage.

9:45am. Things are looking good. AC power up. Router configured. Ethernet cables connected. LCD monitor up, server-laptop up, demo-machine set up with its clear acrylic sides on. Ready to power up. Then I notice that the heat sink for the demo machine has come off and has rolled around the case. Not good. Having relatively heavy metal objects getting up-close-and-personal with motherboard components is bad news.

In the back of my mind I'm calculating the distance to Central Computer (1.0 blocks) and CompUSA (2.5 blocks), and thinking I might have to get a mini-ITX motherboard, pronto.

We re-seat the heat sink on the CPU and fire up the machine. No video. We check cables. We re-seat components. Checking the KVM switch, resetting the LCD... Yes! We're up.

Nothing quite like a morning hardware failure to get the mind in gear.

9:55am. The booth is ready, except for the webcam. I decide it's more important to demo than take time to set up the webcam, and there are two days of show after this one anyway, so we defer it until tomorrow morning.

10:00am. Showtime. The acoustic pitch of the hall changes. It rises from a low hum to a somewhat higher hum, as though someone had gently leaned on an invisible accelerator pedal. Let's see: Tuesday 10-5, Wednesday 10-5, Thursday 10-6. That's 20 hours in the booth. Let's get busy.

People start arriving. This is great! LinuxWorld Expo is happening!

One of the great pleasures and privileges of having exhibited at so many LinuxWorlds is that we get to meet and greet people who have visited us before.

The .ORG Pavilion draws together a special blend of people. Makers and those who use, enjoy, or are curious about our latest work. It is a place where people who enjoy the divine activity of creation gather and celebrate our shared, continuing adventure into places that are at once familiar to us and yet compellingly novel.

We are in the back left corner of the exhibit hall, so the general flow of traffic will take a little time to reach us, though clearly some of our old friends have made a beeline for us. We smile, we greet, we show off our latest work, and we enjoy the pleasure of informally reuniting with those who enjoy our work.

This show we are debuting gPXE, the rewrite of Etherboot we have worked on for over a year. Our meeting with Markus and hpa has given us renewed confidence that we're on the right track, and we've heard our friend Don Becker of Penguin Computing [http://penguincomputing.com/] is looking forward to talking with us. We haven't seen him since the Boston show, and it's always a treat to hear his thoughts. He's not shy about giving his opinion, and in our specialized area an informed opinion is worth a lot, whether it is in agreement or urges caution. Good data is gold.

Also in our booth this show we have hardware from two vendors who have funded development (and thus developers) over the last year.

Solarflare Communications has funded iSCSI development and has loaned us a 64-bit rackmount system (the one Michael drove down to Sunnyvale on Monday to get). With this we will demonstrate our progress in network booting using iSCSI protocol. Solarflare makes intelligent 1 and 10 gigabit ethernet cards that are supported by Etherboot and gPXE. It is rewarding to work with vendors who are advancing the state of the art.

Our other vendor partner this show is Coraid, Inc. which makes AoE (ATA over Ethernet) NAS (network attached storage) devices. While in principle similar to iSCSI, they use a simpler protocol and are lower in cost. They have lent us a networked storage device, and we are demonstrating gPXE's ability to boot directly from this device.

We are particularly pleased about this because gPXE is the only software that can do this, which means one can boot Linux directly from a Coraid NAS box. They funded the development, and Michael wrote the code with testing help from the Etherboot community.

Of course we also have the usual network booting demos designed to introduce network booting to people who have never seen it before. For these we use gPXE and hpa's PXELINUX to load a 2.6 kernel, and LTSP to get to the desktop, if we wish. LTSP is two booths down, and we've had the pleasure of working with them for many years.

Working a booth is an interesting experience, and I recommend it to anyone who wants to quickly understand how people respond to a product or service offering. The questions you get will open your eyes, and you'll soon understand what people want to know.

Network booting, being somewhat esoteric to a lot of people, often requires a bit of explanation. The concept is not new, but because the predominant desktop operating system (Microsoft Windows) is generally not deployed in a thin-client server manner, many people are unaware of the advantages of doing so. That's a good part of what we're here to explain.

10:20am. We've greeted old friends and other .ORG Pavilion exhibitors, and now some of the general crowd is finding our booth. They see our signs, our plexi-clad demo machine, and we smile. Some come up to the booth and say "So, what's this Etherboot thing about", and so we begin.

We point out that the demo machine has no bootable media, but we are going to boot it over the network from our laptop linux server. We explain the advantages of centralized management and lower TCO (total cost of ownership), and how Linux can be used to run 50 instances of OpenOffice from a single server without licensing fees. We push the reset button, and take them step-by-step through the booting process: DHCP, TFTP, PXELINUX, and voila!

As many hundreds or thousands of times we've done the demo, it's still fun to see the light go on in people's faces as they see the possibilities. It's kind of like being a magician, without the prohibition on "telling secrets". In FOSS, the point is to empower as many people as possible. The portal into our communities is not some esoteric bit of knowledge or some secret handshake but simple curiosity, and a willingness to explore.

It's interesting to see how questions drive development. Somewhat after Etherboot was created in the early 1990s, Intel created PXE, which is another way to network boot, using DHCP and TFTP. Until a couple years ago, Etherboot did not support this method. In some ways we were the "not-PXE" people.

In years past, people would come up to our booth and ask, "So, how is this different from PXE?". We would explain that we were first and were based on the same underlying protocols and did similar things, but that answer was unsatisfying.

I think that part of the reason I personally funded PXE development in Etherboot was because I got tired of answering that question at shows. It was a good question. There was no technical reason we couldn't support the PXE specification. We did (and still do) believe that the PXE specification is flawed in some significant respects, but it works well enough to boot machines and has a large installed based in BIOSes and mid-to-high range ethernet cards. So now when people ask, "How is this different from PXE?", we can simply say "We are the Open Source PXE implementation". That is also why the next version of Etherboot will be called gPXE. That way there will be no mistake that our software supports the prevailing specification while extending it in useful ways.

The other question we often get is "How is this different from LTSP?". That's easy to answer. We're all about loading kernels. LTSP is about what happens after the kernel is loaded.

So the demos go on, and we show AoE booting to a lot of interested folks. Robin Miller comes by to say hello. He wants to do a video, but I say I'd prefer to do something tomorrow. We chat excitedly for a moment, and he heads off, saying that he'll be back tomorrow.

Our AoE and iSCSI demos have caught the attention of various vendors on the show floor, and it's nice to see some of them filter over and take a look. We enjoy making our software, and we enjoy working with vendors who will support us in doing so. It's our desire to work collaboratively with as many people as possible, within the bounds of our principles and our preferences, of course.

We have found that listening to people and reaching out, especially to those who might be initially distrustful, often creates positive bonds of trust and cooperation.

11:00am. Our iSCSI boot of Windows 2003 Server doesn't quite work yet. It gets to the final stage of booting, but there is some problem we haven't been able to track down yet. Michael is in back of the booth hacking away in the debugger while I do demos. Anselm has decided that he wants to implement syslog (remote network logging) support in gPXE, which would be a great help in debugging. He is busily hacking away, occasionally asking questions of Michael and me, and sharing demo duty.

11:30am. One of the things I should have done on Monday is to make more of our little 4x5 flyers that we hand out to people. It's a handy memory-jogger for after the show, and it has our website addresses and a picture of our demo machine.

I put a PDF of the flyer 4-up on my keychain flash drive, and Michael takes over demoing as I head back to the hotel business center at the Marriott. They are helpful as usual, and in less than 20 minutes I have 800 little flyers ready to go, copied, cut, and wrapped. Progress.

While I'm out, I take a quick side-trip to Central Computer to get a larger network switch. In our 10x10 booth, we are running a router, five laptops, two rackmounts, and a demo machine. We're out of ports, even with the extra 5-port switch I brought, so I make a quick run over to pick up a 16-port switch.

12:30pm. I'm back at the booth. The crowd is steady, and the Slashdot Lounge is well populated. Michael is busy demoing to some folks interested in AoE. Anselm is deep in thought about UDP packets. All is right with the world.

We hook up the new switch, and keep on demoing. Michael says that we've had good interest in iSCSI booting, and he takes off for a walk around the floor to visit with some vendors who have stopped by our booth.

3:30pm. Lots of old friends come by, lots of demos, lots of laughing. We enjoy chatting with the Debian and Creative Commons folks on either side of us. We share stories and make bad jokes. We remember why we come here year after year. It is worth it every time.

5:00pm. The show lights dim to 50%, a signal that it is time to close for the day. We start to power things down, close things up, and look forward to a good night's sleep. I make sure that all of our power strips are disconnected from the show power, because I have seen morning power-on surges fry transformers and leave people in a very bad way. Chances are slim, but pulling a couple of plugs reduces the chances even more. Finally, we lower the heavy table cloths that we brought to cover our equipment and head for the exit.

I invite Michael and Anselm for some dinner at Mel's Diner just around the corner. We're pretty hungry, as we didn't have breakfast, and my low cholesterol diet is about to be compromised in a serious way. As we are seated in our booth, I know what I crave. I feel like Harold and Kumar must have felt as they finally reached White Castle.

I want a strawberry milkshake. For me, this is a decadence that I reserve for special occasions, because I need to avoid sources of fat and cholesterol in my diet. But tonight, I feel only satisfaction as I consume this nectar of deities.

Anselm's the big eater of our group tonight. Markus was last night, I think. I get the feeling that big portions are popular in Germany. I order the turkey dinner -- I think Michael and Anselm are going for burgers. We all get milkshakes, after they see what mine looks like (and how much I enjoy it).

I find out from Michael that he is on the one-point-five-meal-a-day-on-average diet plan like I am. We just don't get around to eating as there are so many other things we like to do. It's fun to know other people have similar habits.

We finish dinner, pleasantly full, and head back out.

We're all a bit fried, and operating on a variety of time zones in our minds and bodies. I bid adieu to Michael and Anselm, and we head to our separate hotels. We pulled everything together today and made it work. I'm reminded of my favorite baseball announcer, Bob Prince, from when I was growing up in Pittsburgh. He was a great play-by-play guy, and at this moment I'm thinking of how after a really close game, where the Pittsburgh Pirates would pull out a victory, he would laugh knowingly and with complete conviction declare:

"We had 'em all the way!"

(maybe PJ will write that someday about the SCO litigation).

There were a lot of things that could have gone wrong, but they didn't. I really had always felt that we would make it here, and the right things would happen. Being with good people will do that -- give you confidence that you'll succeed.

8:00pm. I'm back in my hotel and have stayed awake about as long as I can. I'm still on east coast time, so I think it's 11:00pm. I fall asleep, pleasantly remembering the day's events, and looking forward to tomorrow.

----- End of Part 3 -----