Tuesday Jan 08, 2008

Working with Fujitsu: Travel to Japan

I work in Sun's Burlington, MA office, about ten miles north of Boston. Three years ago when I was asked if I was willing to work on the APL project, my first question was, "Would I have to travel to Japan?"

Flying to Japan

It's not that I had anything against Japan, or travel in general. But Boston to Japan is a long flight! I hated just having to fly six hours to the West Coast. Depending on the aircraft and the stops, Japan can mean 18 hours on an airplane. The "best" arrangement I've found is flying down to New York (less than an hour), then taking a Boeing 777 direct to Narita International Airport (just outside Tokyo). Fourteen and a half hours in your seat, inside a thin metal tube, above the clouds. You leave Boston on Saturday morning, and with the time difference you're lucky to be in Tokyo for dinner Sunday night.

I did discover that, remarkably, my iPod battery could last almost the entire flight. I also learned that sharing your playlists can be a way to break the ice with the person next to you on a plane. You can see right away that they have an iPod, and no one seems to mind if you say, "Nice iPod! What kind of music do you listen to?" My iPod has a fair mix of oldies like Simon and Garfunkle (is Billy Joel an "oldie" yet?) through current bands like Guster and Barenaked Ladies (are they still "current"?). I was actually surprised that a 13-year-old girl sitting next to me on one flight had many of the same artists on her iPod.

Hotel New Otani and Akasaka

Once at Narita, it's an hour-long train ride into Tokyo on the Narita Express. We'd normally stay at the Hotel New Otani, near the Akasaka district with its many restaurantes and easy access to rail. Whenever a large group from Sun was arriving on different flights from different cities, we'd arrange to meet for dinner on the first night at a Korean barbeque. Since you cook the food yourself right at the table, and this particular restaurant had large tables with several barbeque pits, it could handle a large crowd of wearly travelers, and it didn't matter if you wanted to eat a little or a lot, or came late, or left early to get a good night's sleep. And I think everyone gets a good night's sleep the first night.

The New Otani severed an "American" breakfast buffet. While they serve normal fare like fruit, muffins, and cereal, the two most interesting items at the buffet are french fries, and salad. Thinking about it more, french fries are really nothing but home fries cut the other way. But salad? Of course, after a few days in Japan you realize that most Japanese restaurants don't serve a tossed salad with dinner. Usually by the third day or so, I'd find myself craving lettuce and was more than willing to eat it for breakfast.

The Commute

The main Fujitsu facility we were usually visiting is in Kawasaki, although occasionally we would meet in Kamata. We all traveled around using Japan Rail (JR) or the Tokyo Metro. The rail stations are fairly close to the New Otani, an easy walk in the morning, and a welcome walk after a long day of meetings. Traveling with the rest of the Tokyo commuters was a great (if a bit scary) experience. On my first trip to Tokyo, I momentarily got separated from the group during a transfer. I was quite lucky that I found them again quickly, since I didn't have a clue which train to transfer to, or even what my next stop was. To this day, that common recurring nightmare -- the one where you're back at high school, it's final exams, you failed to attend any of the classes or read the book, you don't have a pencil, and you forgot your pants -- has been replaced by the nightmare where I'm in an unfamiliar JR Rail station, alone, and I can't figure out which train to take back to the hotel.

Of course, getting around Tokyo isn't as bad as my nightmare. Almost all the signs have English subtitles, and most people speak at least a little English. Those that don't speak English are usually friendly and try hard to communicate with hand gestures and charades. I suppose the one thing that bothered me most about the train system was that I didn't have a complete map of the system; I couldn't figure out how to get from point A to point B if it involved changing lines. But unlike Boston with its four T lines (Red, Orange, Blue and Green lines), there are more than a dozen subway and JR Rail lines around Tokyo, and far more stations. I can't imagine the size of a map large enough to encompass the entire system.

On the Metro I was introduced to "pushers". These were very different from the pushers I knew from the New York City subway. These pushers dressed in uniforms, almost like police, and wore white gloves. At the platform, everyone lined up according to marks on the floor, in double file, one line on either side of the subway car door. When the subway train arrived, it stopped exactly on its mark, the doors opened, and a stream of people poured out between the two lines of awaiting travelers. Once the car had expunged its passengers, the two lines of people would stream into the car. As the car became full, the pushers went to work. They would gently, and very politely, push you further into the car, and push the next person in behind you. When the car was nearly fully to overflowing, the pushers would push the next person in, keeping their hand firmly on the back of the last passenger until the doors started to close. Just as the door was about to close on their arm, the pusher would pull their arm out. Not unlike the way my wife packs a suitcase.

As you can imagine, the Metro trains are crammed with people. At one point, two of us missed a stop because we were pushed to the middle of the car, pinned in the crowd, and unable to make our way to the doors for our stop. Of course, I never felt claustrophobic in the subway cars; most of my American coworkers were over 6 feet (1.8m) tall and their heads poked up above the crowd.

I very much appreciated the daily commute. I've done a lot of traveling, and for some cities the only places I've been are the airport, the hotel, and the meeting room. You never get to know a city or its people doing that. But going from the hotel to the meeting room by subway and train, with all of the other commuters, really made me feel like I was in Tokyo, seeing the sides of Tokyo even most tourists don't see.

Meetings with Fujitsu

Meetings were like meetings are. Long days. Small conference rooms. Presentations. Whiteboards. But in this case, they were far more productive because of the time shifting. Normally, Kawasaki is 13 or 14 hours ahead of Boston (depending on daylight savings time). I could send an email today, Fujitsu would read it tonight, and I wouldn't see the response until tomorrow. But when we were all there, in the same room, at the same time, we could discuss things, interactively. We could ask a question, discuss the answer, and come to agreement in a matter of minutes, instead of days.

When you're traveling 10 (from the West Coast) or 18 (from the East Coast) hours to get to Japan, you try to arrange as many meeting topics as you can, to get the most out of the travel time. We'd usually have at least three very long days of meetings on various topics. All in all, trips to Japan were packed with as much work as we could physically accomplish.


After a long day of meetings, it was back to the subway station to head back in to Tokyo. If we were tired, we'd get off the train near the hotel, and eat dinner in the Akasaka district, past the many pachinko parlors to a shabu-shabu or tampura restaurant. Other nights we'd head down to the Ginza. If things were going well, typically on the last night we'd treat ourselves to Kobe beef at a little hole-in-the-wall a couple of blocks off the Ginza called Gyu An (my coworker Roman Zajcew always ate at Gyu An when he was in Tokyo, and traveled to Japan so much that the hostess there thought he lived in Tokyo).

One night our Fujitsu hosts took us to a local dining district in Kamata, and we ate at a small Korean barbeque restraurant with only three tables, and one person who was the hostess, waitress and "chef". A very nice, and authentically Japanese place -- they didn't even have menus in English or with pictures you could point at. It was quite humorous at the end of the meal when both Sun and Fujitsu people pulled out their credit cards and started a friendly argument over who should pay, when the hostess came over and explained they didn't take credit cards; embarassed, we had to pass the hat around the table and collect cash from everyone to cover the check.

After dinner it was usually late and time to head back to the hotel room. Unfortunately, the New Otani had high-speed internet service in the rooms, I always had my laptop, and people in the US were just starting their day. On more than one occasion I was online well past 2am reading and writing emails with my colleages back home, catching up on the issues of the day, and letting them know about the progress we were making. This arrangement also meant that I could email my action items from todays meetings, people back home could work on the tasks while I slept, and I could attend the next day of meetings with several action items already resolved.

Heading Home

There is no better feeling after a business trip than settling in to your seat, hearing the aircraft door close, and feeling the plane back away from the airport. You're on your way home! All I could think about was getting back to my own house, my own bed, and my waiting wife and daughter. My little girl was just two when I made my first trip to Japan; one week away from a two-year-old means missing 1% of her short life. But thanks to my travels, and the little "Hello Kitty" doll I brought home on one trip (which we've named "Edo"), my now-four-year-old has no problem finding Japan on her globe, and knows a few words of Japanese as well.

Monday Jan 07, 2008

Working with Fujitsu: Building Bridges

I was about as surprised as anyone in 2004 when Sun announced the APL agreement -- we were going to partner with Fujitsu to develop the next generation of midrange and high-end servers based on the Olympus CPU chip. I was even more surprised when I was asked to work on the project.

Bridging Timezones

I'm based out of Sun's Burlington, MA office, near Boston. Initially, the I/O team was drawn from Burlington, while the OS and Service Processor teams were from offices in California. Fujitsu is located in Kawasaki, Japan. That's 13 or 14 hours ahead of Boston (depending on daylight savings time). This was going to take some changes to my work habits in order to get some workday overlap with my West Coast colleagues, and attend phone meetings with Fujitsu in Japan.

I used to be an early bird. In my previous job, I'd be at my desk by 7am every day; home for dinner by 5pm. When I joined Sun nine years ago, I found that schedule tough since most people didn't start until 10am, but worked well into the evening. Later when I worked on a project with Sun's West Coast offices, I found the West Coasters also came in at 10am -- 10am Pacific Time! If I started at 7am and quit at 4pm, I only had three hours of overlap with the West Coast (and one of those hours was their lunch time). So my hours started to shift. I started working more 8am to 5pm, then 9am to 6pm.

But now working with Fujitsu, there was more of a timezone challenge. Initially we planned to have monthly face-to-face meetings with Fujitsu to exchange technical information. But soon it became clear that once-a-month meetings were not often enough, and the few people who could travel were not always the right people for the discussions. We had to find a way to meet more regularly, weekly, and involve the right people. Fujitsu generally starts at 9amJT, which is 7pmET.

But the first rule I laid down was: I will not work between 6pm and 9pm. That's family time. If I can't eat dinner with my family, play with my daughter after dinner, and tuck her in to bed every night, then that's not the job I want to have. The rest of the Sun team was very understanding, and we scheduled conference calls at 9pmET/6pmPT/11amJT (during daylight savings time, that changed to 10amJT).

There was still an advantage to being an early bird. Often I would get up at 6:30 or 7am, check email from my home office and find that my colleagues at Fujitsu were still working. We could exchange email and information in real-time. Then when my daughter got up around 8am, I would play with her for a while, then after breakfast, whem my wife and daughter headed out to dance class or swim class or whatever, I'd shower and go back to work, usually around 11am. I'd break for dinner at 6pm, and spend the evening with my family until my daughter was asleep in bed. At 9pm, I'd head back to the home office for a conference call, or just exchange email with the West Coast and Japan. Most work days ended at 10pm or 11pm.

In the end I worked hard, but not as hard as it seemed to others. They'd see me responding to emails at 7am, and still working past 10pm. It gave the illusion I was working 7am to 11pm (I think that motivated everyone else to put in extra hours themselves).

I'm really fortunate that Sun promotes an environment where employees can work from home. While I could have gone into the office every day, I would get there at 11:30 and have to leave at 5:30; hardly worth the traffic and travel time. And since most of my peers were in California or Japan, I would sit alone in my office all day, on the phone with people in other states. Initially, I tried to go into the office three or four days a week. By the end of the project, there were periods of entire months when I didn't see the inside of my office. And the time I did not spend commuting, I spent working. So I think Sun is really fortunate, too.

Bridging Companies

Working with another company to develop a product is difficult. I've done it in the past. But with Fujitsu, we were working with a competitor to develop a product. That's an order of magnitude more difficult.

And Fujitsu was a real competitor at the time. The Sun Fire 6900 and 25000 systems were up against PrimePower in many areas. We needed to work together on the Olympus platforms, but work being done to improve the Sun Fire servers, or pave the way for Rock-based servers, was not part of the contract; sharing that with Fujitsu before SPARC Enterprise platforms shipped meant giving away proprietary information to a competitor. Of course, improvements to support more cores, higher thread counts, larger memory images, and PCI-Express were all related to SPARC Enterprise as well as Sun Fire and Rock platforms. Figuring out what could be shared and what was taboo was a constant chore in the beginning. This was the first project I've worked on, at Sun or anywhere, that we had a lawyer from the Legal department assigned to the project team from day one. Working with another company meant understanding what was OK to share, and what must be kept proprietary.

This of course cut both ways. There were many technical questions we had that the Fujitsu engineers just couldn't answer; they weren't allowed to answer. We at Sun were not used to working like this. If we have a question about a new microprocessor, we're used to walking down the hall and talking to the engineers designing the chip. Working with another company meant trying to make forward progress with limited access to certain information.

Then there were the technical documents that were under Japanese export control -- they could be read in the US, Canada and the UK, but no where else. Sun is a multi-national company, where you're just as likely to have someone on your engineering team from India, Ireland, or elsewhere, and data is usually shared without regard to national borders. Sun largely ignores geography, but on this project, geography became a key factor. I think this was hard at first, but we quickly learned the routine; what was OK and what wasn't. We had to educate others within Sun (for example, the Sun engineer in India who was writing the SunVTS diagnostic tests, but he wasn't allowed to read the technical documents unless he flew to California).

There was also a need for education within Sun about APL. Many Sun employees assumed APL was a co-branding agreement. While this was largely true of the T2000 machines (they were entirely Sun designed, originally Sun Fire T2000 rebranded as SPARC Enterprise T2000), the SPARC Enterprise M-class servers were not simply re-branded PrimePower machines; they were jointly designed, jointly developed, and jointly manufactured by both companies. Fujitsu provided the CPU chips and support ASICs, but Sun was responsible for the M4000 and M5000 hardware design (boards, power supplies, chassis), while Fujitsu owned the M8000 and M9000 hardware design, and both companies contributed equally to the platform-specific Solaris components and service processor firmware that ran across the product line.

Bridging Cultures

It's no secret that Japan has a very different culture from America. Before starting the project, many of us took a short class on doing business in Japan, and we all received a little handbook. But while the cultural differences were present, they were not significant. We Americans tried to be very respectful of the Japanese culture and traditions, but I'm sure we screwed up on many an occassion (I know I did). And in response, the Fujitsu engineers were respectful of the American culture, and forgiving of our faux pas.

We quickly learned the appropriate way to address our peers at Fujitsu, and they us. For example, I would refer to my Japanese counterparts by the last name with the suffix "-san", and they quickly learned that "Bob" was how I preferred to be addressed (and whenever anyone said "Mr. Hueston", I'd jump up and look for my father!). As we got to know our peers better, and were exposed to less formal situations such as dinners or one-on-one phone calls, then using their first names was acceptable. But out of respect, we'd never call someone by their first name in front of their superior.

Language was also a bit of a challenge. In Japan, most children study English in elementary school, so every profressional we dealt with at Fujitsu spoke some English. We always had to be aware that the person we were talking to did not speak English as their first language. We had to speak a little slower, articulate a little more clearly, and try to avoid using colloquialisms. If a sentence wasn't understood and we're asked to repeat ourselves, a natural instinct is to paraphrase what we just said, but that only added confusion. We had to learn to exactly repeat what we had just said, giving our Japanese counterpart the opportunity to re-listen and re-translate the words in their head. If they had a question or needed further explanation, we had to learn to listen carefully to their question, and answer the question that was asked. At first, talking face to face (or worse, phone to phone) was a real challenge, but after just a little while, it became second nature -- you talked more slowly using simpler phrases, and listend more closely to maximize understanding in both directions.

Looking back at the people that Sun selected for this team, I've noticed that they are all people who are calm, respectful individuals. There were no hot-heads, or name callers (Sun has their fair share, but not on this team). Perhaps someone higher up realized that in order to bridge the cultures, we needed a team of people who were patient, and would respect the people and culture of Japan. And I think mutual respect was the one key thing that helped us overcome any cultural differences.

Bridging The Distance

I was going to write about traveling to Japan and meeting with Fujitsu, but I think that can be a posting all on its own.

Building Bridges

Working on the APL program was all about building bridges. In a certain respect, the Sun engineering team was a bridge between Fujitsu and the rest of Sun. We bridged Sun's customers, service engineers, manufacturing engineers, and Sun's business to Fujitsu. In turn, the Fujitsu engineers we dealt with bridged Fujitsu's customers and business interests to Sun.

Tuesday Nov 06, 2007

Sun SPARC Enterprise M-class going strong

In yesterday's earnings announcement, Jonathan Schwartz noted, "We saw particular strength in our high-end systems lineup..." The high-end systems lineup he's referring to is the Sun SPARC Enterprise M-class server line, jointly developed by Sun and Fujitsu. It's a great feeling to work on a product for almost three years, watch as it enters the market, and see incredible demand. And it gives a great sense of pride that the product line is notable in its contribution to revenue and margin.

Since the M-class server line shipped in April, I've been blogging about the neat features and amazing capabilities of these servers. The CPU chip was developed by Fujitsu, but a lot of the usability, servicability and functionality that makes it a great server (not just a great chip) came from Sun, and especially from my software team.

Sadly, most of the M-class software development work is done, and it's time to move on. I'm hoping to help bring other great features to our volume systems and x64 product line in my new role.

I have a couple of posts I've started that still need a little cleanup, but after that, I think I'll be putting this blog in mothballs. Maybe I'll start another blog... The Secrets of Thumper... :-)


Bob Hueston


Top Tags
« July 2016