Sin-Yaw @ Juniper

Companies all have their cadence. After several years, the managers got the hang of it and gained efficiency: ops review, focal, recruiting, budgeting, repeat. Corporate America all have reporting obligations and have long been tuned to the same quarterly rhythm. What amuses an observer is how each company does it differently.

Ostensibly, the budgeting exercise allocates resources for the optimal return. The managerial principles are well-studied: rank all projects according to their NPV (Net Present Value, a conceptual value that represent how much the project is worth today) and fund as many positive ones. In reality, no company does it this way. No method computes NPV accurately; a good manager can always present his/her projects in good light.

The budget cycle, then, becomes an exercise to synchronize the value: a daunting task for any company of decent size. Compared to others, Juniper does it quite well. In few short intense weeks, the list quickly settled. The ripple, of course, has its normal jittering effects. From my vantage point, I don’t see standard bad behaviors: keeping sleuth funds, hoarding resources for political gains, hide projects into skunkworks, using high-priority ones to pad organization, etc. Instead, people got the list, internalized them, and simply moved on to do the works.

Very cool.

Like many companies with a dispersed campus, Juniper provide shuttle services that transport employees from building to building. Hey, losing that hard fought parking space in the early morning is not cool.

This shuttle service also make it possible for me to attend meetings on those days I took the bus to work. I wouldn’t need to beg for rides, just hop onto the shuttle.

So I did. Surprise! A bumper sticker right there. Make my day.

For the most part of my professional career, networking ends with a NIC card. The software stack on top includes a device driver, a chain of software, usually in kernel, that sends packets up and down, and a collection of daemons (programs that belong to no-one and never die) that do their things: ftp, ssh, nfs, etc. I am vaguely aware of one daemon, routed, that figure out which ways to send packets. I really do not care much, since I am more concerned with packets for or from myself. Routing deals with, mostly, packets for others.

When I came to Juniper, networking changed completely. The classic router architecture has three main components: forwarding, routing, and OS.

Forwarding shoves packets from one end to the other as quickly as possible. Juniper’s gears can handle a very impressive amount: probably the fastest in this industry. Huge amount of packets arrive at the interface, each one must be categorized, inspected, and possibly forwarded to the right destination. The machinery operates based on FIB (forwarding information base), a data-structure dynamically updated by the routing intelligence.

Routing is the art of figuring out what the net looks like by what the neighbors said. It generate RIB (routing information base, used to be called routing table) that is the foundation of FIB. When wrongly computed, the net melts down or critically disrupted. If done right, companies can deliver more services without spending more: a key competitive advantage for many enterprises these days.

The OS manages the machinery and all those sophisticated software. It provisions resources for payloads and provides a stable platform for innovations. In my book, it serves the most important function: allowing routing and forwarding to innovate independently.

Ostensibly, the box looks the same as a standard data-center server. There is power-supply, lots of fans blowing air around, a chasis that connects boards, called blades these days, that plug into it. Off the back of the box, many cables run off to a far-away corner. I walk into a networking lab thinking how similar it is to a data-center.

This is probably why an OS guy like me feels so at home in a networking company.

How do you make a large number of people to change at the same time?

Note the operating words. How implies a methodology that can be learned and practiced. Make signifies the intent. Large so that this is non-trivial. Lastly, same time means the degree of control.

Military perfected this art long time ago. Recruits go through a process that strips individuality and imprints discipline and obedience. After that, rigid structures guarantee the clarity and efficiency of the communication. When the commander gives the order, a massive number of people turn on a dime, at least when they are well-trained.

The fashion industry has a different approach. They have mastered human’s innate tension between novelty and conformity. With clever messaging, the industry seduces society to change voluntarily, quickly, and predictably.

Companies change too, some successfully, like GE, Intel, HP, and several others commonly studied and written about. Countries sometime change at stunning speed. Look at the transformation of China. No one visited 10 years ago would recognize its current state.

But how? To effect changes for a large number of people, the change agent must first understand those very people and execute with a methodology. People change twice: once internally and the 2nd time in behavior with their group. In the beginning of the process, people look for reasons not to change. Once the tipping point has been reached, they embrace it and the change accelerates. The challenge becomes keeping the balance of fast proliferation and the lost of control.

Of course, there is the traditional corporate-style change management: get senior executive endorsement, socialize with stakeholders, build early successes, prepare for communication, etc. We are in the era of blurred line between employees and community. New change management must embrace both approaches: the tipping point style and the corporate style.

A manager well-versed on both change styles will be rich rewarded in his/her career path.

The bus fare is $1.75 one way, gasoline costs about $3 per gallon, the vehicle costs about 50 cents per mile to operate, in terms of wear and tear, how long the commute must be to justify riding bus, or not driving? Or what will be the MPG (miles per gallon) threshold of the car to make it work?

If your car is a Prius (45 MPG) and your commute is longer than 3 miles, you save money. If an SUV (15 MPG), 2.5 miles will do. If we ignore the vehicle’s wear and tear, it takes 26 miles for Prius and 8.75 miles for the SUV.

EcoPass reduces the fare to zero and takes economy out of the consideration. Good job.

Cross posted

You have everything planned. Strategy is brilliant, wheels are turning, projects are in motion, future is unrolling. You feel good and can almost taste success. Unexpectedly, something big happened: two big buildings collapsed, a hurricane wreaked havoc, an earthquake shook a romote place, something called sub-prime is disturbing the US financial institutes. The event has no direct effect on what you are doing, but everyone is talking about it. Should you change your course or stand firm to ride out the storm? Since there is no data to support either decision, it is essentially a gut call, wrenchingly.

Guts are quite a rare commodity these days.

The question is whether the event has fundamentally changed the course of the world, or it is merely a ripple to be forgotten.
Sequoia advised their portfolio companies to change. In fact, to jump immediately.

If the event altered the landscape. Companies that reacted quickly also recovered quicker. Companies that tried to ride it out experienced painful cut-backs and frequently never recovered. There are many examples. Studious readers have already listed both columns. The million dollar question is, “How soon will this economy recover?”

You know Sequoia’s answer by now, “Not soon enough for you to just do nothing.”

Juniper reminds of China all the time: energetic, optimistic, impatient, and creative: precious like a talented young adult ready to change the world. All what’s needed is a good tool chest and some practicing.

The timing is everything. Choose a good direction and the world changes according to the plan. A bad habit may hinder competitiveness beyond remedy. I am very fortunate to join the company at this time. The opportunity to participate an era does not come frequently. This, again, reminds me of China too.

I met many people and made friends. Breaking-in is always scary; Juniper made it easy. Several senior executives went to length to introduce me to other leaders. I became part of a company-wide tiger team. What followed was 6 weeks of whirl-wind accelerated assimilation.

In the midst of those, I was still moving back from Beijing to my west San Jose old home. I stole a trip back to Beijing to wrap up 3 years of life there. On my way back, I took a detour, too short, to visit Juniper’s Bangalore site. There I got a renewed appreciation of my 3 years as the remote guy. I came back mentally excited, physically tired, and over-whelmed by a garage full of moving boxes.

Juniper soon entered its annual planning cycle. Every companies does this. Juniper’s process is a familiar ones: messages exchanges at various levels until they are reasonably consistent. This is just like routing: adjacent nodes exchange information until their view of the world become consistent. Only the physical layer here is the budget.

I felt sure-footed enough to start trying out some change ideas. They coincided well with the second phase of the tiger-team project earlier. These new activities kept my daily calendar booked wall-to-wall almost everyday. The bus rides, 40 minutes one-way, gave me precious time to collect thoughts and calm-down from the adrenaline-filled days. I am not sure what will I do when winter comes.

100-day is just a calendrical event. Today will roll like yesterday or tomorrow. In my heart, this marks a milestone of becoming a part of this company. Many more friends to come, I am sure.

I found myself preaching. Maybe I do it all the time, but rarely I was self-aware of that fact. I must have delivered the same sermon enough times already.

What separate professional software development and amateur? The ratio of effort devoted on quality. Amateurs make it work. Professional show evidences. The amount of efforts in professional software houses devoted to quality and verification easily dwarfs standard definition of development effort.

The verification takes place in phases. First is the effort to make sure the software works as designed. This is most commonly done by the same person who designs the software, but more effective if by someone who sits close to him/her and watches over his/her shoulders as he/she codes. The goal is to make sure the software behaves well for all ranges of inputs.

Next is the effort to verify that software meets the requirements. Most likely, someone who understands the requirements will come up with methods to verify such, without giving too much concerns on how the software was designed.

After these two steps, the developer confidently declares his/her software works. He/she now must prove the new does not break the old. This is simple enough. Just run all those previously developed tests in the first 2 steps.

Imagine many of such teams following the same process. When they all come together, they do not work together. There could be misunderstanding of the interfaces or inadvertent changes that affect other parties. This is why we need integration tests.

At the end, companies frequently will “try run” on their best customers’ environments. Sometimes, the best effort “in the labs” does not discover problems “in the real.” This is the attempt to be as “real” as possible. It also ensure the satisfaction of the best customers.

Some requirements are hard to verify. Specifically, performance requirements demand benchmarking, reliability needs stress testing, security needs attacks. None of these come “naturally,” and software houses frequently use simulators. They create un-natural but concentrated faults to put the software under duress. The most common simulator is for new hardware platforms that are still in development.

This whole system is under the supervision of a general test harness that automates everything, supplemented with as little manual tasks as possible, for maximum execution speed. This harness handles test sequencing, failure management, metrics of progress, report generation, and even provisioning of resources.

People asked why do I come to Juniper. “Because the way I watch TV now.”

Just a few months ago, I routinely fired up BitTorrent at night to download episodes of my favorite programs (House, Heroes, Grey’s Anatomy). I had no choice. Those programs are not aired in China. When I moved back to the US, I don’t wait anymore. I watch those programs from the official web sites: fox.com, nbc.com, etc. I will hook up my laptop to the new 52″ 1080p HDTV. Many programs play in full-screen mode (Fox is better) and I will soon forget that’s not live TV. I laughed so hard on Tina Fey’s SNL performance on my laptop, displayed on my TV set, Sunday afternoon — no need to stay up Saturday night.

Internet never busted. Many Internet companies went under, but the net kept on growing. Billions of people around the world are waiting to get online. Internet these days are several times larger than the bubble days and sees no sign of slowing down.

And which companies are to ride the wave? The one that produce the biggest, meanest, and fastest routers for the mass. There are other players, but Juniper is at the center of this stage.

Juniper is large enough to make a difference and small enough to experience explosive growth. Juniper’s executives are all on-board for this vision. Everyone focuses on the customers, the market, the revenue, the products: not on internal politics.

When I watched SNL, I noticed the page said more than a million viewings and counting. This is main-street behavior now. It takes lots of networking to deliver millions of viewings.

Can’t think of another company that is readier.

I have lived and worked in silicon valley for decades. Except for a few train rides to Palo Alto, I have never taken public transportation to work. Unlike Tokyo, New York, or even Beijing, there is no good network of mass transporation here. There has never options other than driving. Then I started working for Juniper Networks.

My work location is right next to a bus hub. Route 26! Isn’t that the one that run through the street 10 minutes from home?

A simple Google later, I am ready. On a nice day, I left the house with my computer backpack and walked toward the bus stop. Walk! 10 minutes later, I stood waiting at the bus stop with exact change in my pocket. I felt nervous: would the bus show up, have I missed it already, what kind of people would I found onboard, could I deal with the bus?

I went onboard, paid the fare, and faced a sparsely occupied bus. I sat down at the back corner and surveyed. My fellow passengers are all minding their own business, catching Zs, absorbed into the MP3 player, or busy studying the air. They are all clean commuters. Just like myself.

Juniper is the last stop. So I leisurely disembark. For the journey, it covered about 10 miles, costed $1.75, I left home at 7:45, arrived the stop at 7:55, the bus came at 8:05, I walked into the office at 8:50. I enjoyed Art Pepper Meets The Rhythm Section all the way. iPod is a wonderful thing.

Later, I found the company will give out EcoPass: a sticker for free VTA buses and light-rail. Sweet.