Sin-Yaw @ Juniper

When I started this job in July, Juniper did not have an active blogging community. That changed drastically since, and I am part of that transformation. I found myself questioning the purpose of this one.

If Juniper employees will read my internal blogs, who read my external ones? Why don’t they just go to Loud Thoughts? I cannot come up with a compelling reason. In fact, my lack of activity here answered that question.

So, readers, I am retiring this blog. If you are Juniper employees, click Core Dump to read my recent posts (I share this blog with Mike Bushong and others). Juniper employee or not, you are always welcome in Loud Thoughts.

Vendors, suppliers, companies, whoever heed these get my money.

• An HDTV that works with my computers. I am a computer person and use it as my DVR. I connect my computer to the TV in a painful way. I would like their relationship to be similar to the one between iTune and iPod, the TV being the iPod. The physical connection should be simply BlueTooth, USB, or Ethernet. Don’t create new connectors.

  I don’t want another computer in my living room Gateway2000-style. I just want my TV to be computer friendly or network savvy.

• Downloadable music and video without DRM. Online contents should enjoy the same rights as traditional media. If I buy a CD or DVD, I can rip it into my computer, give or lend it to my friends, and use it on as many devices as I please. If I pay for a song or movie online, why can’t I have the same rights?

  On similar note, I want CDs and DVDs to be priced as if they are selling in China or India. I can buy perfectly legal “Asia exclusive editions” for a fraction of the costs. They play equally well for my iPod. The world, after all, is flat.

• I want all portable devices to use the same charger, preferably without any connector. I would like to throw all my portable devices on the coffee table and find them fully charged the next morning. This technology exists. Can’t be that hard.

  As an acceptable secondary option, please all use the small USB connector.

• Hulu that caches. With today’s fatter pipe, I still get buffering pauses for a full-length movie. I also want to watch them when there is no connection. The point of Hulu is for me to watch those commercials, having a copy locally should not matter.

• No subscription. I want NetFlix, Wall Street Journal, New Yorker, Economist, etc. sold per issue. Offer me the same options available in the print world. I can buy an issue from the stand; I can also subscribe.

  Same thing with the hot-spot wireless services. Once in a while, a better connection, or the only one, requires me to pay. I am OK with paying for them. A better service should be rewarded with my money.
  I just want the experience of purchasing two hours of internet connection similar to buying a cup of coffee or tank of gasoline. Here is my credit card. Thank you. Done..

This Internet thing is good at satiating curiosity. I started a pursuit this weekend to get a grip on the industry, defined as: Cisco, Alcatel, Juniper, HuaWei, F5, Extreme, RiverBed, and Arris Group. (Foundry and Nortel are kind of hard to get these days.) I am interested only in their financial performance, so I dug up their latest federal reports, or whatever their websites will give.

Relative to each others, Cisco dominates with 51% market share. Alcatel follows with 25.7%, HuaWei 15.5%, Juniper 4.7%, and the rest combined is 3.1%. Interesting that each major sales region has a dominant player: US, Europe, and Asia Pacific. The numbers show Juniper is winning share, but I can be biased.

If economy of scale plays in this industry, smaller ones — Arris, F5, Extreme, and Riverbed — will face consolidation pressure. Since this is high-tech, I am not very sure economy of scale really play.

Among the “big 4″ (Cisco, Alcatel, HuaWei, Juniper), Juniper has the best gross margin (67.5% of total revenue) and HuaWei least (33.9%). This is not surprising that Juniper is at the “high performance” segment and HuaWei competes on costs. (Alcatel is 34%, just a hair better than HuaWei.) Juniper spends the most on R&D at 20.7% and HuaWei the least at 11.4%. HuaWei has the lowest OPEX (27%) and Alcatel highest (54.5%). Cisco makes the most money with 20% net income and Alcatel lost at -20.2%.

Alcatel deserves some attention with low gross margin and high OPEX. Not a good combination: products cost too much to make, company is not efficient. Over the other end, Riverbed has a whopping 73.5% gross margin and equally high OPEX of 74.8%. On the surface, this means they make and spend money both quickly.

Juniper’s and Huawei’s numbers make sense. One invested on R&D and the other kept the corporate overhead low. Extreme is interestingly pretty much on average in every areas. Mediocracy can be difficult to rid.

Yes, today. I was recruited to encourage people to leave comments on blogs, particularly Juniper employees.

Many of you sent me email messages or stopped by to tell me your thoughts on my blogs. One of you even convinced me to take an entry down. (Actually, it was two of you.) As a blogger, I frequently wonder, “Why don’t you just leave a comment.” The typical answer was, “Oh, I just want to tell you about it.”

But we bloggers love comments. All bloggers are basically self-publishing writers. Writers love readers. Writers want confirmation that they stirred, stimulated, touched, amused, entertained, influenced, or, when it was a really good day, inspired someone.

So, leave a comment on someone’s blog today. It makes their day.

Thank you.

There is a simple and classic strategy to maximize the chances of winning: spread the opponent’s resources wide and pierce his weak point with concentration of your strength.

Why would Cisco start a server business? I guess that someone in Cisco found out that routers are really the same, from manufacturing’s and engineering’s points of view, as servers. If they make good routers, they should be able to make reasonable servers too.

But server and router businesses are quite different. The components of servers are mostly commodity these days. To differentiate, Dell claims the cost leadership, IBM and HP built armies of service personnel and a vast array of system software. Sun has Java and Solaris.

It will take years and billions of dollars for Cisco to build up this business and compete with IBM, HP, and Dell. All of them very good at raging attrition wars against smaller players. It does not make Kevin Johnson happier when John Chamber make enemies with Mark Hurd, Samuel Palmisano, and Michael Dell at the same time.

Cisco did not announce a sofware strategy, they simply mentioned vmware. The implication is that OS is up to the customer and run on top of the virtualization layer. We all knew that Cisco is not very good at producing a strong operating system (most of the variations of IOS are not full-feature OS). This seems like a good spot for Juniper to poke.

2009 should be a good year for Juniper.

I optimized the route to building 1: go down the stairway on the gym side, cross street, enter building 1, turn left to the corridor, up the elevators, 4th floor. This is my engineering instrinct: must optimize for heavy usage.

A new device appeared several weeks ago. It is an over-sized ashtray. I was puzzled.

How much money Juniper spent on this apparatus? In addition to the giant ashtray, there is also a sign telling the smokers to use it. Prior to the new device, smokers simply use the trash can that is 6 inches away.

The dooor enterinng building 2 has a sign, “No smoking within 25 feet of the door.” That new ashtray just barely 25 feet away, or slightly less. Hmm. Maybe one must smoke beyond the sign-post? But that’s on the road. Is that a hint?

That ashtray also the starting point of my routine jogging routine. I will be a bit annoyed to meet a herd of smokers at the end of my run. Good thing that never happened.

Since building 2 houses the cafeteria, this smokers’ station will also get maximum notice with the heavy foot traffic. Guess that’s why few smokers use it. Most of them simply hover somewhere in the parking lot near the main entrances. They tend to avoid paparazzis these days.

I am not sure there are more such smoker’s stations around Juniper campus. I am thinking of request it be moved, then that appears to be a “not in my backyard” petition. Oh well.

slide:ology The Art and Science of Creating Great Presentations Nancy Duarte ISBN: 9780596522346 O’Reilly Media, Inc. August 12, 2008

Completely reasonable for a graphic design company to publish a book on slide design or presentation style. The principles are completely dead-on: spend the time to prepare; watch out 3 elements: delivery, visual, and message; focus on the presenter not the material; and use contrast to emphasize.

The professional approach, however, makes the book less a tutorial than an advertisement. Ver few people can afford 90 hours for preparation, creating custom diagram, establish an image library, or do professional color schemes. Not all presenters will be Al Gore. In fact, many who present frequently are stretched in both time and money to just get by with slides slammed together quickly and a few minutes of mental preparation. They are the ones who need helps.

Gartner, a market analyst, publishes its Magic Quadrant reports for various industries periodically. Earlier, it released the one for enterprise firewall. I was very pleased to see that Juniper is at the leaders quadrant. We have the right vision and can execute to secure future wins.

By the way, I met the co-author John Pescatore several years ago to discuss some security topics. This man has lots of gems to offer to anyone seeking his wisdom.

The cautionary notes confirmed my decision, in previous life, to discontinue an enterprise firewall product. Only the strong can play in this tough market. All network infrastructure vendors, however, must offer network security solutions. Juniper, therefore, is a “pure-play” vendor here. I am glad to see our favorable position against Cisco, our competitor since the company inception.

How do you find out what are truly valued in an organization? Everyone who has been with the organization knew its values implicitly, but rarely can anyone articulate them.

I decided to stack rank my organization several months ago. I, however, did not know how to. Should I rank them by their IQ, EQ, or the average of them? What about leadership, teamwork, communication, or other soft skills? Does Juniper value engineering prowess above all else? Or, like some companies, verbal eloquence?

I asked a team to decide the categories of evaluating individuals. I also asked them to weigh each category. Four of them worked 2 months on this assignment. They talked to many in my organization and discussed the topic at length.

Drum roll…

35% productivity, 15% creativity, 25% people skills, and 25% dependability.

Ella, the team lead, showed the similarity to the Juniper Way. High-Quality results come from productivity. Agreement is strongly linked to people skills. The dependability is the cornerstone for relationship. In addition, engineers always value creativity.

We then ask every managers to score their employees according to this system. There were intensive discussions on the differences of each other’s scoring style, but almost none on the categories, or the weights.

Now I have a consistent performance model across my entire organization and a common language to communicate. This system will reward people for their productivity, people skills, dependability, and creativities. More importantly, it rewards those who are most balanced across all 4 categories, instead of those who are unproportionally strong in one, but weak in others.

I emphasized that the system is designed to be 85% accurate in assessing employee performance. It is a non-goal to be more accurate; doing so risk replacing judgment with processes. This process met two objectives: I now have a concrete value-system that is consistent with the Juniper Way; I also have a consistent and quick methodology to identify the best talents in my org.

I anticipate this process to evolve as managers work on it. For the 1st time deployment at Juniper, it surpassed my expectation.

Thank you. Ella & team.

Any reasonably large organization must manage employee performance — separating the wheats from the chaff — rewarding those who brought in value and aligning talents to tasks. This has been practiced widely. It comes down to a simple concept: bucketization. Putting talents in buckets of various labels, exceptional, very strong, strong, etc., and deploy different incentives, or disincentives, accordingly.

If this is not controversial, why not nano-bucketization: putting almost everyone in his/her own bucket? Somehow, we are comfortable with calling 20% of the population exceptional, but not two of them “the best two” and the next person “the 2nd best.”

Think about the objectives of the performance process. Wouldn’t it make sense to be more precise? The precision, it seems, actually hurts teamwork. The destructive competitiveness among individuals somehow diminishes when the classification is sufficiently coarse. We are the best team, but each team member contributed equally and should not be treated differently. This is understandable from the team’s point of view, but not managerially. The manager needs to know the relative strength of each member.

Hence the paradoxical conclusion: managers must stack rank his/her organization, but not communicate the result with precision. Every study that shows the negative outcome of stack ranking misses this point. Stack ranking is absolutely necessary. It is done implicitly by every manager anyway. The communication of the rank, however, should probably never happen.