Creative Engineering

A presentation at Parallel Data Lab Annual Retreat in October 2010 in Pittsburgh, PA, USA by erik riedel

Slide 1

Slide 1

Crea%ve  Engineering   PDL  Dis%nguished  Alumni  Talk   Erik  Riedel,  PhD     October  2010   Bedford  Springs,  PA  

Slide 2

Slide 2

Slide 3

Slide 3

Crea%vity  

Slide 4

Slide 4

  Exposure  to  diverse  fields  and  diverse   ideas  is  cri%cal  to  be  able  to  see  the   “big  picture”  when  faced  with  a   problem  in  your  own  domain.  

Slide 5

Slide 5

Conferences  among   my  trip  reports   Magazines  on  my   nightstand   • • • • • • • • Economist   Business  Week   Wired   Fast  Company   Technology  Review   Harvard  Business  Review   Forbes   Talk*   • • • • • InfoVis   CODES   Int’l  CES   Emergence   SOUPS   • ISCA   • ASPLOS   PLUS  –  any  book  with  “innova%on”  in  the  %tle   *no  longer  in  print,  Tina  Brown  interview  magazine  

Slide 6

Slide 6

  The  same  thing  is  true  in  discussion   among  your  peers  and  colleagues.     Inclusion  of  diverse  ideas  and   approaches  leads  to  superior  solu3ons.  

Slide 7

Slide 7

Seagate  Ac%ve  Storage  Tour  

Slide 8

Slide 8

  Most  important  advice  –  get  out  of  the   office  regularly.     Whether  to  conferences,  or  to  talks  or   to  visit  other  divisions  in  your  company,   or  (best)  to  visit  your  customers                   (the  target  users  of  your  technology).  

Slide 9

Slide 9

Consumer  

Slide 10

Slide 10

Slide 11

Slide 11

Enterprise  

  Go  into  someone’s   data  center  and   look  around   Picture  from  flickr/ewwhite  

Slide 12

Slide 12

  Look  in  the  corners  too   Pictures  from  flickr/ewwhite,  craig1black  

Slide 13

Slide 13

Passion  

Slide 14

Slide 14

  Enthusiasm.  

Slide 15

Slide 15

  Results  ma\er.  

Slide 16

Slide 16

There  are  always  more  solu%ons  than   barriers,  but  there  sure  are  a  lot  of   barriers….    real  and  imagined       Budgets,  skills,  headcount,  informa%on,   freedom-­‐of-­‐ac%on,  poli%cs,  processes  

Slide 17

Slide 17

  Don’t  mis-­‐underes%mate  the  power  of   ge_ng  your  hands  dirty.     Prototypes,  systema%c  analysis,   evalua%on.     Make  the  %me  for  rigor.  

Slide 18

Slide 18

Engineering  

Slide 19

Slide 19

  There  are  a  lot  of  ways  for  an  engineer  / scien%st  to  make  an  impact.*           *  The  difference  between  a  scien3st  and  an   engineer  is  that  the  engineer  stops  to  think   how  much  it  will  cost.  

Slide 20

Slide 20

Pa%ence  

Slide 21

Slide 21

Objects   Ideas   • U.S.  6,298,401  Anderson   • CMU  NASD   – Network-­‐A\ached  Secure  Disks   • NSIC  NASD   – Wide  support  &  feedback   • T10  OSD  (SNIA/ANSI)   – Wide  support  &  feedback   – Worked  out  lots  of  details   • Storage  Bricks  (Gray)   – Saw  the  win,  spread  the  word   Results   • Lustre   • Panasas   • IETF  pNFS   • EMC  Atmos   Influences   • Google  GFS   • Spinnaker   • S3   • ….  

Slide 22

Slide 22

Slide 23

Slide 23

Slide 24

Slide 24

Pa%ence  II  

Slide 25

Slide 25

Slide 26

Slide 26

See  U.S.  Patent  7,464,106     (January  2006)  

Slide 27

Slide 27

Slide 28

Slide 28

Slide 29

Slide 29

Patents   • 6,732,241    Technique  for  migra%ng  data   between  storage  devices  for  reduced  power   consump%on   • 7,200,747    System  for  ensuring  data  privacy   and  user  differen%a%on  in  a  distributed  file   system   • 7,405,941  Storage  array  with  enhanced  RVI  

Slide 30

Slide 30

Slide 31

Slide 31

Influence   • Research  vs.  Product  Groups     • CTOs  have  an  impact  almost  exclusively  by   influencing  the  “real”  parts  of  the  business   – Product  Management   – Sales  &  Marke%ng   – Business  GMs  –  caveat:  don’t  confuse  power  with  influence  

Slide 32

Slide 32

Communica%on  

Slide 33

Slide 33

Dinners   Mee%ngs   Mee%ngs   Career   Progression   Code   Amount  of  %me  spent  

  *Advice  courtesy  Jim  Gray,   as  retold  in  May  2007  at     UC  Berkeley  on  a  day  both   sad  and  upliping  

Slide 34

Slide 34

  Never  underes%mate  the  power  of  talking  and   listening.               PowerPoint  is  way  over-­‐rated.  

Slide 35

Slide 35

  There  are  numerous  studies  that  show  that  your   personal  network  is  the  single  biggest   determinant  of  career  success.     These  studies  are  correct.  

Slide 36

Slide 36

Summary   Crea%vity   Passion   Results  &  Rigor   Pa%ence   Influence   Communica%on  

Slide 37

Slide 37

References   • Listen:  Discover  the  Secret  to  GeIng  Through  to   Absolutely  Anyone  by  Mark  Goulston   • The  Tipping  Point:  How  LiLle  Things  Can  Make  a   Big  Difference  by  Malcolm  Gladwell     • Made  to  S3ck:  Why  Some  Ideas  Survive  and   Others  Die  by  Chip  &  Dan  Heath   • Design  Thinking  (IDEO  and  others)   • The  Future  of  Enterprise  IT  by  Geoff  Moore  at   2010  SNIA  CloudBurst  event   • Gobioff/Riedel  Emigra%on  talks  –  startups  vs.  labs  

Slide 38

Slide 38

What  wouldn’t  fit  into  the  talk   OTHER  THOUGHTS  

Slide 39

Slide 39

Cloud  Compu%ng  

Slide 40

Slide 40

The  previously  separate  roles  of  sopware  developer   and  opera%ons  have  [become]  increasingly   intermeshed  and  intertwined.  Things  are  materially   different…   Ray  Ozzie,  Chief  SoUware  Architect,  MicrosoU   Programmers   Ruby/Rails,  MapReduce   IT  Managers   VM  images,  vApps,  VLANs  

Slide 41

Slide 41

  Cloud  is  open  an  “excuse”  for  enterprises  to   move  to  “New  IT”  –  away  from  the  old  client/ server  model  that  has  been  used  for  the  past   ten  years  [toward  Web  2.0  IT]   Werner  Vogels,  CTO,  Amazon  

Slide 42

Slide 42

Consumer  vs.  Enterprise  

Slide 43

Slide 43

  The  way  we  run  our  lives  has  forever  changed.     The  employees  we  are  hiring  right  out  of  school   are  appalled  by  the  technology  we  use  to  run   our  companies.  They  are  more  produc%ve  at   home  than  they  are  in  the  office.     Marc  Benioff,  CEO,  Salesforce.com  

Slide 44

Slide 44

From  September  2010,  SNIA  CloudBurst  keynote  by  Geoffrey  Moore  

Slide 45

Slide 45

Consumer  A\en%on     • Diagram  from  Maya   Gap   Courtesy  Mick  McManus,  MAYA  Design  

Slide 46

Slide 46

Small  Nodes  vs.  Big  Nodes  

Slide 47

Slide 47

From  April  1999,  Ac%ve  Disks  talk