HPE and Juniper

 Several readers have asked: "What do you think of the HPE / Juniper merger completed last month?

And yes, our nephew's wife has worked at HP and now HPE, in networking, for two decades.  Plus I've known Tom Black from his early HP days to his Cisco days to his Aruba days and HPE days, before his shift to HPE Storage (https://blocksandfiles.com/2020/07/28/hpe-storage-takes-the-black/) in 2020.  Tom is currently on sabbatical, and out of the networking game for awhile now.   HPE's Aruba division is currently at a $5 billion annual run-rate.

The Big Kahuna, of course, is Cisco.   I did have the privilege of doing 150+ interviews of early Cisco folk for the Cisco Foundation a few years ago, which was a tremendous opportunity.  AND, I did get the 'award' from Cisco for being 'the guy who made Cisco successful' from none other than the EVP of Engineering, Joel Bion, when I was hosted to talk about Ray Price's and my book, The HP Phenomenon, in 2010.  Boy, did that surprise me!  I briefly thought, "Well, why didn't I get any founder's stock?"   

That story turned out to be interesting.  Cisco couldn't get any traction in the Local Area Networking milieu in the early to mid-1980's.  Their gig was Wide Area Networking, connecting Networks of Networks.  But the VCs, some 70+ of them, turned Cisco down for years.   If there aren't yet many LANs, what is the point of building a  WAN to connect some non-existent LANs?

Thanks to a serendipitous connection to Cisco from Bert Raphael in Corporate Engineering, and a project that he and Tony Fanning led to tie all of HP's far-flung engineering operations 'together' via computer teleconferencing (https://dl.acm.org/doi/pdf/10.1145/637069.637109), HP became the leading commercial purchaser of Cisco routers in 1985, with Mark Laubach and his college mentor, David Farber, serving as lead designers, along with Hank Taylor in HP's IT shop.   Laubach is now at Ciena, another modestly successful networking company.

No, let me put it more strongly about the HP and Cisco synergy--HP was the ONLY corporate purchaser of Cisco routers for 28 months, from May 1985 until September 1987, when Boeing bought one, partially due to Dave Packard's recommendation, while he was on Boeing's Board of Directors.   And I was the person who signed the original Cisco Purchase Order for HP, with co-founders Len Bosack and Sandy Lerner sitting in my conference room at the HP Deer Creek facility. 

And you might have read the Sandy Lerner article that I wrote in 2016 for the Computer History Museum (https://computerhistory.org/wp-content/uploads/2020/01/core-2016.pdf).  Her story is an intriguing one, notable for a number of reasons which we'll skip here. 

 It is also perhaps worth noting that a remarkable man, unfortuantely a paraplegic, Jim Pelkey, conducted an amazing set of 81 interviews of the key networking founders and leaders in 20+ companies and 12 other organizations and institutions during the 1986-1988 period.   He omitted Cisco, because they didn't yet amount to anything (his view). These were audio interviews that the Computer History Museum eschewed in 2012, until I weighed in to say "THIS IS THE HISTORY".  Amazing how myopic and prejudicial, groups can be on occasion.   I flew to Hawaii to interview Pelkey, and to review his voluminous materials.  Then we got some contributed money, and a true historian (Andy Russell), and they've now all been transcribed and put on line at CHM, and there is even an ACM book (Circuits, Packets, and Protocols https://www.amazon.com/Circuits-Packets-Protocols-Entrepreneurs-Communications/dp/1450397271) that details much of this.

All this prelude is to say that I've had occasional small glimpses of the networking world, including ACM interviews with Bob Metcalfe (3Com), and a partnership with Bill Krause (3Com) along the way, plus Andy Bechtolsheim about Arista, and . . . . So, the point of all of that dialogue is to say that I've had some fascinating opportunities around the networking world.

Okay, so what's the scoop on HPE with Juniper, and how does that stack up with Cisco and Arista?

Well, first of all, Juniper was a sterling challenger to Cisco at the turn of the century, growing spectacularly from $4M in 1998 to $675M in 2001, with a market share of 33% in routers, dinging Cisco's early dominance dramatically.   Billed as a David and Goliath face-off, the two fought hard battles over the next decade, with acquisitions and new product claims and all sorts of hoopla.  Juniper revenues, though, plateaued after 2010, roughly at $5 Billion per year (Figure 1), while Cisco has not done much better (Figure 2).  Cisco, of course, is ten times bigger, clearing $56.6 Billion in the past 12 months (to August 2025).   

By contrast, newcomers, especially Arista (founded by Andy Bechtolsheim in 2004 after he sold another company to Cisco), is managed by seasoned network executives (mostly from Cisco), has grown nicely, particularly after introducing the software-defined networking products.  These have been around for two decades, but until Arista solved the access speed problem, they were not much in favor.   Now, the clear technology winner is software-defined network solutions, and their run-rate at $8 billion (past 12 months) is 50% larger than either Juniper, or HPE Aruba.  Surprise!   And the only significant annual growth-rate for all of these companies is Arista, at a stunning 31% per year compound rate for the past 12 years.

Here is a synopsis of the various definitional differences for Cisco, HPE/Juniper, and Arista.   Ciena had a nice early contribution, but in truth Ciena lacks momentum at this point.

Another way to illustrate these comparative growth rates is to plot the quarterly gains or losses vs. the same period a year prior.  Figure 5 shows this for the four networking companies described, with the key difference that the HPE diagram is for all of HPE rather than just the Juniper networking acquisition.

Okay, so HPE will roughly double its current networking run-rate revenue with the merging of the Juniper product lines.  That will put it slightly ahead of Arista's recent results, but the momentum for the HPE combination is zero, while Arista's is 30%+ per year.   In addition, there will be inevitable thrash at HPE, as the Aruba and Juniper offerings are merged and sorted out.

Which stock would you bet on?

 

Another "Oldtimer" exits--Carl Cottrell

 It is always a bit jarring--finding an old friend and colleague in the obituaries.   Carl Cottrell, a true veteran of the "old HP" made the Los Altos Town Crier (and later the San Jose Mercury-News) two weeks ago: https://www.losaltosonline.com/people/obituaries/carl-cottrell/article_41df88ba-a240-41bf-b6e4-5612f8df345a.html.

Carl was a consummate salesman, who rose to some amazing heights for HP along the way.  He spent 39 years at HP, which he chronicled beautifully in his abbreviated biography for the HP Archives:   https://www.hpmemoryproject.org/timeline/carl_cottrell/carl_cottrell_memoir.htm#part_14.   This archive entree covers his first 18 years at HP, starting in 1952.   His memoir covers lots of the early names and contributors, from the customer support group and the quality assurance group through the emergence of formal marketing efforts and, of course, the sales organizations.

His first "big break'" was being asked to manage HPSA, the European sales organization, which he started on January 1, 1961, and he did that well, quadrupling sales in four years.  He was then asked to move to the Eastern region and consolidate all of the disparate teams that had just been acquired from rep organizations, a thankless and tedious task.  He enlisted Bob MacVeety as his junior partner, a man I got to know extremely well when Bill Terry brought Bob to HP Colorado Springs as our Marketing Manager.  That's a story for another time--Bob and I crossed swords several times, but he put together the HP Logic Analyzer belt buckle program, which was the finest incentive for sales people ever conducted!

Sales folk earned a Bronze buckle by selling ten Logic Analyzer units; a Silver for 25, and a Gold for 50.  You got a set, including a Platinum one, for 100.   People went nuts for these, from 1974 through 1982.  MacVeety had been HP's most successful salesman at Bell Labs for years prior.

Cottrell, back in the States, was on hand for the November 1966 launch of the HP 2116A Computer.  As he wrote: "We were fortunate in that we had a few salesmen working in Paramus (HPs New Jersey office) who had computer experience.  Listening to them, MacVeety and I recognized early on that computers could not be sold like instruments.  Much more effort would be required.  Customers wanted to see the computer run applications that they needed before they would buy.  They needed programming instruction and they needed the appropriate peripherals to go with their computer.   Urged on by this realization, we invested in HP's first Data Center in the Paramus office and we equipped it with demo 2116 systems.   Soon there was a steady stream of customers visiting our center.  Based on our Paramus experience, we began to add salesmen in the Region who had computer experience and could 'speak the language.'  Soon, Eastern Region was selling more computers than any other Region, and we had set the standard for others to know.  I was proud of our Region."

Cottrell's memoir had just one more line: "Satisfied with my accomplishments in Eastern Sales Region, I brought my family back to Palo Alto to take on other assignments."

There Cottrell's recorded story ends, at least for the HP Archive.   But in fact the later saga is fascinating. 

HP leadership had no understanding of just how tough this computer business might become.   In all of 1967, HP sold just 5 units of the HP2116, all from Eastern Sales, all by Dick Slocum, a salesman working for Cottrell.  1968 was better, and Eastern Sales led again.  Hewlett brought Cottrell to Palo Alto  to work with the design teams.  His observations were trenchant:  "They really were different.  The lab was full of bearded, sandal types.  Quite a shock to somebody who's not used to it.  It was quite a different language spoken there.  We had people to whom the HP philosophy was just totally strange, but we had very, very sharp people.   The challenge was to rein them in.   (Tom) Perkins was running the computer dividison, Bill Davidow was marketing manager, and Jim Treybig (Tandem founder later) was working for us.   Many people left and became stars."

Then, as Cottrell later told it, in 1969, the computer business was burgeoning, up 500% at a time that Packard had left for Washington and Hewlett was now HP's CEO and President.  And Hewlett had a particularly hard time with Perkins, whose minicomputer line now was $30 million per year, and Tom had much bigger plans.   "Perkins kept pressing, and Hewlett, piqued, brought Carl Cottrell to Palo Alto to manage a new Computer Systems Group, with two divisions (Dymec, run by Perkins, and AMD run by Jerry Carlson)." It was a near-total disaster.   Carlson, an ambitious accountant, had no qualification for managing much of anything, but Packard had long insisted that 'good people could learn any job.'   Perkins, who would go on to co-found Tandem and then the remarkable Venture Capital firm, Kleiner-Perkins, was more than well-qualified, but often incredibly prickly as a colleague.   He soon left HP.

Cottrell, with John Doyle's help, established low-cost overseas divisions, especially for core memory and keyboards, but Data General and Digital Equipment were being run by aggressive, take-no-prisoner types, and Cottrell's and his team struggled, especially after Perkins deplaned.   Cottrell, much later, observed ruefully that, "I gave it my best shot, but we made some mistakes, and I was over my head, to tell you the truth."

A very positive side of Packard's management approach was to take 'failed' executives from one assignment, and put them into challenging new roles that maybe weren't quite so demanding, but could take advantage of the learning that they had experienced.   Cort Van Rensselaer had failed miserably as the Oscilloscope division manager, and then also as head of corporation acquisitions, but he and Cottrell teamed up to define and design an Information Architecture diagram for HP in the mid-1970s, and in key areas defined the entire MRP and later Supply Chain Management systems that gave HP an incredible headstart on running a large multinational corporation with highly effective structures.   HP never chose to productize this work, but you all know Salesforce where Marc Benioff installed these concepts.

This in retrospect, was a huge contribution for HP.   As noted in The HP Phenomenon (p. 188   https://www.sup.org/books/business/hp-phenomenon),  "we documented and then started describing for customers what happens via information flows--invoicing, order processing, manufacturing schedules.  It is amazing--HP pioneered accounting standards--it was unheard of in those days, but HP people had long since set up accounting standards, so that every division had the same chart of accounts.  And we were closing our books in record time."

John Young later gave Van Rensselaer and Cottrell credit for helping, more than anyone else, OR EVEN THE PRODUCTS THEMSELVES, HP being able to become successful in the computing world.  This frankly is one of those behind-the-scenes stories that seldom gets told, but is the secret often.   Young, with Cort and Carl, enabled by computer sales teams that hosted CEO and C-suite workshops in Cupertino (this was a key assignment for Scott Futryk, who became my partner at AstroVirtual many years later), gave this a clear voice in a 2005 interview with Ray Price and Chuck House:

"I was trying to figure out how to run a whole company, particularly applying this quality defect idea in all phases of the company.  We rolled that idea up, and began having seminars here in Cupertino, to which we'd invite a lot of high-powered people to come.  We had a lot of great data.  That was one of the most powerful selling tools ever . . . because we had zero credibility in big corporate IT places.  These were IBM-owned--every company, outside of the lab or the factory floor.  If you're going to break into where the real money is in the enterprise, you've got to go up against IBM.  You couldn't go sell a Spectrum, you couldn't sell a product.  You're selling an answer to a business problem, so you have to tailor the presentation to the corporate communications and inventory guys and corporate manufacturing.  Cort van Rensselaer, Carl Cottrell, Hank Taylor, and Sally Dudley had credibility.   On a daily basis, HP could get sales from around the world--product by product, office by office, the next morning.  The notion that you could get that in 1975 was crazy for any other company.  We had the order processing system, we had the order transmission system, and the email that layered on that.  It was the leading-edge worldwide system.  It was remarkable."  And, it propelled HP from the 14th largest computer company in the world to the #1 position in a mere 32 years.  Yes, without Spectrum and HPPA, this would not have happened.  Yes, without major screw-ups at DEC and IBM circa the 1990 debacles, this would not have happened.   But, without question, without Cort and Carl, it wouldn't have either.

And that story doesn't make the obituary, where Carl's family modestly noted only that "he worked for 39 years at Hewlett-Packard, which enabled him to travel extensively and live abroad.   Many will remember Carl for his contagious sense of humor, which left a lasting impact on many lives."

Carl was an inspiration for me, a person and colleague whom I will always treasure.

Joel Birnbaum

 Barney Oliver, long in the tooth, was set to retire in late 1980 as HP's only CTO or VP of R&D.  While he had shepherded HP's successful entry into "Personal Computing" via first the HP 9100A Desktop Calculator (an incredible marvel for its day, one that computer historians eschew as being a 'PC' for purist reasons) and then via the stunning scientific calculators (HP 35, followed by a whole string of 'firsts'), Barney was agnostic at best about HP entering the mainframe / minicomputer business.  True, he did garner a number of new patents for his novel switching power supply designs for HP's 2000 family, but that had little to do with computing per se.

And truth of the matter be known, HP not only was a laggard in computing circles with old architectures, outdated machines, and a plethora of 'families' which did not interconnect (e.g. compilers for one would not work with another, nor were peripheral drivers consistent, etc.).   It was more and more a quagmire.

John Young had recently become the first non-founder CEO, chosen over his fiery colleague Paul Ely by David Packard as 'the responsible one.'   Ely had flogged the computer divisions much as HP leaders had always done, to be the best at their 'little fiefdom' whether it be CPUs, disc drives, or recorders.   This translated into NIH and general unwillingness to work with other divisions to assure that individual produdcts would work together.  The result was much duplication of relatively simple projects in order to have coverage for the disparate 'computer systems' that HP produced (e.g. HP 1000, HP 2000, HP 3000 and HP 9000 families), which were all independently derived and supported lines, each needing a full panoply of support hardware, software, peripherals, and ancillary equipments.   It was maddening for customers, and incredibly expensive for the development teams with very little advantage from unique contribution.

What to do about it?

The signal was a big one, not just a red flag but a RED FLAG OMEN.  Stanford University sponsored an architectural summit around this new idea from IBM, called Reduced Instruction Set Computing.   And, in a more than irksome fashion, HP was not invited to send anyone to the summit.   Hewlett and Packard at the time were the largest benefactors to Stanford in its entire history; John Young was a Trustee, and John Doyle was intimately connected to the school.   Moreover, HP had helped initiate, underwrite, and had become the flagship user of the Stanford Honors Co-operative program for working engineers to advance their education while working.   (Note: I was already one of the beneficiaries of this program, earning an MSEE from Stanford while working full-time at HP).

Young and team, chastened by this affront, met to discuss "What next?" and their conclusion was that they needed a strong computer systems leader to step into Barney's shoes, at least for the compute side.

Who?

This was a plum job for leading headhunters, and Joel Birnbaum, IBM's Computer Science leader, was the chief target, especially since his lead researcher, John Cocke, had led the first RISC design.  When approached, however, Birnbaum was diffident at best.   His observation--HP in computing was about one-twenty-fifth the size of IBM, and came in at about number fourteen in annual compute revenues.  Why would he bother with a job at such an also-ran?

Fifteen years at IBM had not created a great fortune for Birnbaum, but measured in terms of 'the team' it was a plush job.  For his Computer History Museum interview years later, he recalled some of his IBM teammates fondly, including 16 IBM Fellows and a Nobel Laureate.   The list he quickly recited as colleagues also included 11 Turing Award winners and 6 Eckert-Mauchly award winners from the prestigious ACM computer society.

HP had (and still has) no historic Turing winners, Stanford has nine.   HP did have two Eckert-Mauchly winners in recent years (Norm Jouppi and Josh Fisher, both earning their awards before their HP time).

After several entreaties, though, Birnbaum accepted an invitation to visit (he told an interviewer that he decided he could get a free trip for him and his wife to the Bay Area so that she could attend an opera in San Francisco).   As unforeseen events sometimes unfold, the net result was that HP was able to persuade him to join, as the replacement for Oliver's computer research leader.   This story, lengthy in its detail, is contained in The HP Phenomenon, pp. 265-273 (https://www.sup.org/books/business/hp-phenomenon)

A fascinating vignette in that book is when Birnbaum told two IBM colleagues that he was going to visit HP, and John Cocke said: "He's' going to give you the Barney Oliver interview."   Birnbaum asked, "what is that?" and Cocke replied, "Get ready to take another PhD exam."   Birnbaum said to Cocke: "Bullshit.  He's not going to give me any PhD exam."

Birnbaum went on in the CHM interview to say that he walked in to meet a scowling Oliver, arms folded, who commanded in a Jehovah-like booming voice, "Sit down."  Oliver then, according to Birnbaum, "went through all the pleasantries that he could handle, which was maybe eleven seconds."

When I heard this description from Birnbaum, it instantly took me back to my own first interview with Barney almost twenty years prior--and his contempt for my lack of full understanding of the three small "C's" of calculus--conformal mapping, convolution integrals, and celestial mechanics (tools that were in extremely modest usage around HP laboratories).

Well, net net--Birnbaum offered that RISC designs could unify the disparate HP computer offerings, and HP leadership leaped on the idea, and forthwith, he came.  To run the HP Computer Research program.

But HP had more activity than computers underway.  It was a $2.5 billion annual revenue company, with just $1 billion in computing.   Most of the revenue, and most of the engineers in fact, were involved with scientific instrumentation -- for engineers of all stripes (aeronautical, mechanical, electronic, software), and scientists (chemists, biologists, astronomers, physicists, mathematicians, and environmental science).

Rather than saddle Birnbaum with all this, HP put John Doyle in place as the overall VP of R&D, with Joel reporting to him.   And then, in an unexpected move, Doyle and Young concluded that the basic R&D health of HP, especially the creativity side, had been ebbing with the constant emphasis on low-hanging fruit for short-term revenue improvement rather than longer-term breakthrough products (welcome to the Harvard Business Review management style--e.g. the last decade or two demise of HPE, Cisco, Boeing, and today's unfolding tragic story of Intel).   

My great advantage was that I was ready for a change, and I'd had some surprising success with two kinds of HP projects--first, stand-alone products: the large-screen displays that had irked Packard for a time, and then the highly successful leadership line of logic state analyzers that truly dented the old Tektronix armor; and secondly, "integrated systems" evidenced by the Microcomputer Development Systems (HP 64000 family), which decimated Intel's early leadership with chip developers.  This latter  program required integrating technology from 14 HP divisions to make a coherent system, something the computer divisions had singularly proven unable to do.  See the HP Journal, 32 pages, October 1980.   https://hparchive.com/Journals/HPJ-1980-10.pdf

So, Young and Doyle invited me to create HP's Corporate Engineering department to emphasize both increased innovation and increased productivity via co-ordinated development.  Tall order, and I won't here claim wild success, but it had the great advantage of letting me have the "D" side of Barney's job, while Birnbaum got the "R" side mostly (and Doyle picked up some "R" for instruments as well).

Thus, when someone sends a note about "Who was Birnbaum?" it causes me to reflect and just be thankful that I was privileged to get to know this incredible man more than most.

I'll add more about Joel and what he did for HP and for the world in a subsequent post.

Barney Oliver

 Odd questions sometimes arrive.   This one today was about Joel Birnbaum, viz the Computer History Museum blog post in the InnovaScapes blog.   Who, exactly, was Birnbaum, and where did he fit into the HP story?  The story has two parts, the first of which requires describing Barney Oliver and the lab he constructed for HP.   The second part is what do you do when a genius of the Oliver type retires after thirty years, for a company trying to transform itself from an instrument to a computer company?

It goes back, way back, to Hewlett and Packard at Stanford, where a much younger Bernard M. Oliver was enrolled, and became good friends with Hewlett and acquainted with Packard.   He graduated from Stanford at seventeen, and from Caltech with a PhD by 21.  Working at Bell Labs with Nobel Laureate Claude Shannon and Marconi prize winner John Pierce, he co-developed Pulse Code Modulation and laid the groundwork for satellite communications.  By 1952, at age 36, Hewlett persuaded him to leave Bell Labs and run research and development for a fledgling HP.   

Barney commented on the wide disparity in the two companies: "At Bell Labs, I became quite used to a very high, uniform technical level.  The lab at HP was assembled catch-as-catch-can, out of people who happened by. . .". On the other hand, "the pace was much slower at Bell, . . . you didn't feel the blowtorch on you to get the job finished to nearly the extent I felt it when I first came to HP."

I heard Barney give a talk at Caltech in April 1955, talking about the advent of satellite communications, two and a half years before Sputnik dazzled the world.  I was an impressionable fourteen years old, mesmerized by Barney's energy, enthusiasm and obvious brilliance, with no idea what a Hewlett-Packard company was.  But I did know about Bell Labs.

In July 1962, I was ushered into Barney's office at HP, on my first day working for the company.  He famously interviewed (and approved) every new engineer for HP, but he'd been out of town when I interviewed, so I was hired without his stamp of approval.  Since HP in Palo Alto (virtually the entire company at that time) only hired three engineers in 1962 (vs. eight in 1961), it was a small task.

Oliver had just joined the Palo Alto School Board, and he was intent upon changing the curriculum to "the new math."  He queried me about my math classes at Caltech, asking about the "three small C's" of calculus.   Many of you know about the "big three" -- Differential, Integral, and Vector Calculus".  Not so many know about the "small three" -- Conformal Mapping, Convolution Integrals, and Celestial Mechanics.   While I had taken them, I quickly found that I was nowhere near as proficient as he was.

I had won a Science Fair award years earlier for a color television exhibit (10 years before color TVs were to be sold in America), and based on that, plus my mother's birth in Colorado, Oliver assinged me to the Oscilloscope division which was scheduled to move to Colorado in a year or so, saying that it was HP's only 'failing' division, and they needed someone creative to help them overcome the key competitor Tektronix.  

His inducement was "this is HP's only division that provides a two-dimensional answer (e.g. an X-Y plot) rather than a scalar number.  You might do something with those displays."

Five years later, with Oliver's grudging tacit agreement, I bucked David Packard's dictum to cease work on a large-screen X-Y display, and we introduced the first commercially viable high-speed X-Y display for computer graphics.   This earned Packard's opprobrium for some time, but after a mere 18 years, he awarded me the "Medal of Defiance" for 'extraordinary contempt and defiance' of the CEO's direct order, and he mentioned it in his autobiography, saying "how do you handle insubordination when the guy is right?"

Barney was a true force for HP development.  Carver Mead summed it up nicely for the Franklin Institute, noting that while he got great credit for the first PC (the HP 9100A, by some definitions) along with the HP handheld scientific calculators,  "he is the one who blew the bullshit out of the HP design labs."   Many rued his strident, abusive style, but no one doubted his talent.

Barney's career at HP was outstanding, but he did other extraordinary things as well.   He was IEEE President along the way (1965).  Bill Hewlett had been President in 1954 of IRE, the predecessor group to IEEE.   He held 42 patents at Bell Labs, and another 25 at HP.   

Oliver calculated the re-entry requirements for the Space Shuttle tiles, he contributed to nuclear energy designs, and he helped perfect the safety net for the Bay Area Rapid Transit trains (BART).   Perhaps his greatest legacy was pioneering SETI, the Search for Extra-Terrestrial Intelligence.   Here's SETI's bio for him.   https://www.seti.org/people/bernard-oliver/

One way to measure his impact is to realize that at his death in 1995, aged 79, he was one of only six Americans who had been inducted into both the National Academy of Engineering and the National Academy of Science.   

What, you may be asking, does this have to do with Joel Birnbaum.  The answer is centered around the question of "how do you fill Barney's shoes when he retires?"   And that gives us the basis for the next post, which is quite a story in itself.