The TANDBERG Way of Engineering

Tandberg Telecom, also known as just TANDBERG - the videoconferencing company - was a very successful organization started from scratch around 1990 and then, during the next two decades, effectively competing and systematically outperforming all other market players in the videoconferencing industry. By 2010, TANDBERG was a billion dollar business with a 50% marketshare and 1700 employees worldwide. In April 2010, TANDBERG was aquired by the american tech behemoth Cisco Systems for 19 billion norwegian kroners (3.3 billion USD).



TANDBERG (1990-2010) was a legendary success that deserves to be praised and studied. TANDBERG was characterized by a unique organization culture, a fantastic sales force, very competent and charismatic leadership, but perhaps first of all, it had an engineering team that was able to churn out spectacular and successful products.

by Olve Maudal, a 60 minute talk at NDC TechTown, Kongsberg, September 4, 2019

Transcript

1. Tandberg Telecom, also known as just TANDBERG - the videoconferencing company - was a very successful organization started from scratch around 1990 and then, during the next two decades, effectively competing and systematically outperforming all other market players in the videoconferencing industry. By 2010, TANDBERG was a billion dollar business with a 50% marketshare and 1700 employees worldwide. In April 2010, TANDBERG was aquired by the american tech behemoth Cisco Systems for 19 billion norwegian kroners (3.3 billion USD).
 
 TANDBERG (1990-2010) was a legendary success that deserves to be praised and studied. TANDBERG was characterized by a unique organization culture, a fantastic sales force, very competent and charismatic leadership, but perhaps first of all, it had an engineering team that was able to churn out spectacular and successful products. by Olve Maudal, a 60 minute talk at NDC TechTown, Kongsberg, September 4, 2019 The Way of engineering
2. • History of TANDBERG (1990-2010) • Example of a product - The C90 (aka Saturn) • Observations from TANDBERG • Principles for effective product development • Q&A
3. • History of TANDBERG (1990-2010) • Example of a product - The C90 (aka Saturn) • Observations from TANDBERG • Principles for effective product development • Q&A
4. Tandberg Display Tandberg Storage Tandberg Romfart Tandberg Data Tandberg Carwash & Pizza
5. + = ca 1990 Jan Chr Opsahl Videofonprosjektet
6. TANDBERG’s first proper product (ca 1991) 3
7. 1993 1995 1997 1995 1997 1997 TANDBERG (1993-1999)
8. 2000 2002 2001 2002 TANDBERG (2000-2008) 2004 2004 2006 2006 2004 2004
9. 2008 2009 2009 2009 TANDBERG (2008-2010) 2010 2009 2009
10. 0 1 2 3 4 0 250 500 750 1000 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 Revenue (MUSD)
11. 0 125 250 375 500 0 250 500 750 1000 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 (the numbers are approximately right) Revenue (MUSD) # Engineers
12. 0 125 250 375 500 0 250 500 750 1000 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 (the numbers are approximately right) Revenue (MUSD) # Engineers
14. #2 in 2008 #1 in 2009 #1 in 2010
15. Presseklipp 19. April 2010 TANDBERG solgt for 19 mrd NOK til amerikanske Cisco Systems
18. • History of TANDBERG (1990-2010) • Example of a product - The C90 (aka Saturn) • Observations from TANDBERG • Principles for effective product development • Q&A
19. TANDBERG Codec C90 (aka Saturn)
22. HD input H.264 H.264 H.264 H.264 HD output H.264 H.264 H.264 H.264 HD input HD output
23. HD input H.264 H.264 H.264 H.264 HD output H.264 H.264 H.264 H.264 HD input HD output
24. • 1 Altera Cyclone III 120 for Audio switching (Nios II softcore 50 MHz) • 9 TI 6727, audio dsp for echo control, compression, decompression, +++ • PowerPC 8347, main processor, application software, networking, user interface • 3543 components / 15659 pins • 16 layers • 3264 nets C90 MAIN BOARD
25. • 10 Da Vinci DM6467 for video compression/ decompresion(1 ARM, 1 dsp, 2 coprocessors), • 5 Altera Cyclone III 120 for video scaling & composing(Nios II softcore 50 MHz) • 15 Gbps video backplane • 3.8 GByte DDR2 RAM • 128 mbit x5 SDRAM • 6097 components • 30520 pins • 22 layers • 6490 nets C90 VIDEO BOARD
26. •analog amplification •high quality AD and DA converters •pure electronics, no processor/SW •717 components •6 layers C90 AUDIO EXTENSION BOARD
27. • 10000+ components • 44 (6+22+16) layers • 56 processor cores • several million lines of code (C and C++) C90 - from a geek point of view
28. • Developed at Lysaker • Started spring 2007 • First HW prototype arrived summer 2008 • Released late 2008 (~20 months of development) • 2-3 people working with mechanics/design • 4-5 people working with electronics/hardware • 5-6 people working with FPGA development • 40-50 people working with software development • 4-5 test developers • 1 person working with approvals TANDBERG Codec C90
29. • Continuous planning • Always attack high risks first • Heavy focus on effective feedback mechanisms • Visualization of actual status throughout project • Teams: GUI, App, Protocol, Video, Audio, FPGA, Platform, QA, Support • Parallel development • Iterations and time-boxing • Daily 15 minute morning assembly of elders • Weekly rendezvous meetings • Early and many prototypes Development Practices in the Saturn project
31. March 2010 “For most of my life I wanted a Porsche, now I think I want a Tandberg EX90 instead.” Wainhouse Research
33. PC based solutions
34. Networking products
35. And a lot of other stuff
37. • History of TANDBERG (1990-2010) • Example of a product - The C90 (aka Saturn) • Observations from TANDBERG • Principles for effective product development • Q&A
38. Development Process?
39. Development Process?
40. As an engineer joining TANDBERG...
41. •No documentation •No routines •Fooling around •Not following plans •Decision are postponed •Nobody decides •Little respect for management •Little modularization •Lack of precision •Sloppiness •People not working hard for many, this was the first impression...
42. •No documentation •No routines •Fooling around •Not following plans •Decision are postponed •Nobody decides •Little respect for management •Little modularization •Lack of precision •Sloppiness •People not working hard but then you started to notice... •People communicate •Focus on important stuff •Embedded slack •Continuous planning •Effective decisions •Autonomous organisation •Respect for doers •No integration period •Spectacular products •Deliver early •Sustainable pace
43. •No documentation •No routines •Fooling around •Not following plans •Decision are postponed •Nobody decides •Little respect for management •Little modularization •Lack of precision •Sloppiness •People not working hard •People communicate •Focus on important stuff •Embedded slack •Continuous planning •Effective decisions •Autonomous organisation •Respect for doers •No integration period •Spectacular products •Deliver early •Sustainable pace ... and while you still saw the "negative" stuff, you started to appreciate the "positive" stuff even more.
44. • People communicate • Focus on important stuff • Embedded slack • Continuous planning • Effective decisions • Autonomous organisation • Respect for doers • No integration period • Spectacular products • Deliver early • Sustainable pace Observations from TANDBERG Agile? Lean? Deming? Scrum?
45. • People communicate • Focus on important stuff • Embedded slack • Continuous planning • Effective decisions • Autonomous organisation • Respect for doers • No integration period • Spectacular products • Deliver early • Sustainable pace Observations from TANDBERG
47. Negative and Positive Components
48. everything has a negative component ...
49. ... as well as a positive component
50. so if you want to improve something...
51. ... do not just try to fix the negative stuff ...
52. ... without also understanding how it will affect the positive component ? ? ? ? ?
53. because you might end up by reducing the positive component by even more
54. mediocre organizations often have a profile like this great organizations might have profiles that looks like this “Managing your problems can only make you good, whereas building your opportunities is the only way to become great.” (Good to Great, Collins, 2001) !
55. • History of TANDBERG (1990-2010) • Example of a product - The C90 (aka Saturn) • Observations from TANDBERG • Principles for effective product development • Q&A
56. Few high tech projects are like running down on a paved road where you can see the ...
57. ... goal in the end of the road.
58. Typical projects are more like...
59. extreme orienteering
60. in difficult terrain
61. with a group of people
62. in the dark
63. with only a sketchy map as guidance
64. and you should all end up at the cabin... together!
66. Embrace chaos
67. Celebrate the organization (not individuals, nor teams)
68. Admire the doers (and create an autonomous organization)
69. Focus on communication (over documentation)
70. Introduce slack If your company’s goal is to become fast, responsive, and agile, more efficiency is not the answer--you need more slack. (Tom DeMarco)
71. Beware the observer effect
72. Use constraints to set direction “Det kan du bare drite i” (no f&*^* way) Per Haug Kogstad (arguably the most influencial person and key driver behind TANDBERGs success.)
73. Break the rules “Man skal bryte regler!” (you should break rules!) fuck you!
74. Reward courage (and failures)
75. Focus on the whole product system thinking vs reductionism
76. Everyone has a plan 'till they get punched in the mouth. – Mike Tyson Continuous planning Plans are of little importance, but planning is essential – Winston Churchill Plans are nothing; planning is everything. – Dwight D. Eisenhower No battle plan survives contact with the enemy. – Helmuth von Moltke the Elder
77. Aim for approximately right rather than accurately wrong
78. Release early and release often
79. Seek early feedback Pixar - our films are never finished, they are just released
80. Never reveal your next products
81. Avoid corporate standards and procedures
82. Focus on execution, forget about innovation initiatives
83. Collective ownership
84. No time writing or detailed cost control "What we don't do is treat our employees like they're all, you know, criminals," (Jenn Mann, SAS Institute)
85. Build the company on trust
86. Hire for passion, attitude and skills
87. • Embrace chaos • Celebrate the organization, not individuals • Admire the doers • Focus on communication • Introduce slack • Beware the observer effect • Use constraints to set direction • Break the rules • Reward courage • Focus on the whole product • Continuous planning • Aim for approximately right • Release early, release often • Seek early feedback • Never reveal your next products • Avoid corporate standards • Focus on execution • Collective ownership • No time writing or detailed cost control • Build the company on trust • Hire for passion, attitude and skills The Way of engineering
88. !
89. • History of TANDBERG (1990-2010) • Example of a product - The C90 (aka Saturn) • Observations from TANDBERG • Principles for effective product development • Epilogue • Q&A
90. Did Cisco destroy it all?
91. No! (of course not)
92. The engineering team at Lysaker is stronger than ever
93. The current product portfolio is beyond spectacular
95. Unboxing and Logistics
96. Machine Learning and Artificial Intelligene
97. Snorre Kjesbu (the key driver for successfully transferring TANDBERG into the Cisco machinery... and vice versa...)
99. Corporate Spring Prox Dynamics Avizia Media Network Services Pexip Surface Dynamics Electric White Skiwo Seevia Huddly Acano Synergy Sky Videonor Vaion Jazz Networks Videxio Starleaf PiOctave Minuendo Spotics Oivi Vaion NNNN
100. Other questions?
101. !
102. “Jeg var arbeidsløs og startet med en tom ølkasse” (Jan Christian Opsahl) «Jeg fungerte ikke under Gerhard Heiberg. Ledelsesfilosofien i Aker er at sjefen tar alle beslutninger i all sin visdom. Deretter utfører slavene ordre.» «Vi vingler fra grøftekant til grøftekant, og når vi så beslutter, har vi en så jævlig fart at vi treffer med 180 km i timen. Men da er vi også fremme før de andre har startet.» «Hvis vi leverer 100 prosent kvalitet, så går vi konkurs. Vi må gå inn for 99,99 prosent, det er omtrent hva vi har råd til. Den siste 0,1 prosenten blir rett og slett for dyr. I elektronikk må det være rom for feilmargin.» AUTONOMY MOMENTUM APPROXIMATELY RIGHT