• Save
Netflix in the cloud 2011
Upcoming SlideShare
Loading in...5
×
 

Netflix in the cloud 2011

on

  • 22,070 views

Talk given at QCon Tokyo April 2011

Talk given at QCon Tokyo April 2011

Statistics

Views

Total Views
22,070
Views on SlideShare
21,790
Embed Views
280

Actions

Likes
52
Downloads
423
Comments
2

20 Embeds 280

http://localhost:4000 118
http://www.cloud24by7.com 45
http://cleonte.github.com 23
http://s23a121.dm1.oii.oki.co.jp 17
http://www.clubcloud.org 15
http://hernangarcia.com 12
https://www.rebelmouse.com 11
http://xacdo.net 10
http://blog.sahsu.mobi 7
http://twitter.com 5
http://www.hanrss.com 3
https://twitter.com 3
http://www.xacdo.net 2
http://www.linkedin.com 2
https://www.linkedin.com 2
http://fmitsuaki.blogspot.com 1
http://localhost 1
http://paper.li 1
http://posterous.com 1
http://a0.twimg.com 1
More...

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
  • Very attractive; watch video record at: http://www.infoq.com/presentations/Netflix-in-the-Cloud
    Are you sure you want to
    Your message goes here
    Processing…
  • it could be used as a real case for a Youtube alike startup.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Netflix in the cloud 2011 Netflix in the cloud 2011 Presentation Transcript

  • Ne#lix  Cloud  Architecture   Qcon  Tokyo  April  12,  2011   Adrian  Cockcro<   @adrianco  #ne#lixcloud  h?p://slideshare.net/adrianco   acockcro<@ne#lix.com  
  • Who,  Why,  What   Ne#lix  in  the  Cloud   Cloud  Challenges  and  Learnings  Systems  and  OperaJons  Architecture    
  • Ne#lix  Inc.   With  more  than  20  million  subscribers  in  the  United   States  and  Canada,  Ne9lix,  Inc.  is  the  world’s  leading   Internet  subscripAon  service  for  enjoying  movies  and   TV  shows.     InternaAonal  Expansion   We  plan  to  expand  into  an  addiAonal  market  in  the   second  half  of  2011…  If  the  second  market  meets  our   expectaAons…  we  will  conAnue  to  invest  and  expand   aggressively  in  2012.  Source:  h?p://ir.ne#lix.com  
  • Unlimited  streaming  for  $7.99/month,  large  and  growing  catalog  of  movies  and  TV  
  • Adrian  Cockcro<  •  Director,  Architecture  for  Cloud  Systems,  Ne#lix  Inc.   –  Previously  Director  for  PersonalizaJon  Pla#orm  •  DisJnguished  Availability  Engineer,  eBay  Inc.  2004-­‐7   –  Founding  member  of  eBay  Research  Labs  •  DisJnguished  Engineer,  Sun  Microsystems  Inc.  1988-­‐2004   –  2003-­‐4  Chief  Architect  High  Performance  Technical  CompuJng   –  2001  Author:  Capacity  Planning  for  Web  Services   –  1999  Author:  Resource  Management   –  1995  &  1998  Author:  Sun  Performance  and  Tuning   –  1996  Japanese  EdiJon  of  Sun  Performance  and  Tuning   •   SPARC  &  Solaris ( )  
  • Why  is  Ne#lix  Talking  about   Cloud?  
  • Ne#lix  is  Path-­‐finding   The  Cloud  ecosystem  is  evolving  very  fast  Share  with  and  learn  from  the  cloud  community  
  • We  want  to  use  clouds,   not  build  them   Cloud  technology  should  be  a  commodity  Public  cloud  and  open  source  for  agility  and  scale  
  • Why  Use  Cloud?     For  Be?er  Business  Agility  For  Unpredictable  Business  Growth  
  • Data  Center   Ne#lix  could  not   build  new   datacenters  fast   enough   Capacity  growth  is  acceleraJng,  unpredictable   Product  launch  spikes  -­‐  iPhone,  Wii,  PS3,  XBox  
  • 20  Million  Customers   2010-­‐Q3  year/year  +52%  Total  and  +145%  Streaming   25   20   15   10   5   0   2009Q2  2009Q3   2009Q4   2010Q1   2010Q2   2010Q3   2010Q4  Source:  h?p://ir.ne#lix.com  
  • Out-­‐Growing  Data  Center   h?p://techblog.ne#lix.com/2011/02/redesigning-­‐ne#lix-­‐api.html   37x  Growth  Jan   2010-­‐Jan  2011  Datacenter  Capacity  
  • Ne#lix.com  is  now  ~100%  Cloud   Account  sign-­‐up  is  currently  being  moved  to  cloud   All  internaJonal  product  will  be  cloud  based   USA  specific  logisJcs  remains  in  the  Datacenter    
  • Leverage  AWS  Scale   “the  biggest  public  cloud”   AWS  investment  in  tooling  and  automaJon  Use  many  AWS  zones  for  high  availability,  scalability   AWS  skills  are  most  common  on  resumes…  
  • Leverage  AWS  Feature  Set   “the  market  leader”  EC2,  S3,  SDB,  SQS,  EBS,  EMR,  ELB,  ASG,  IAM,  RDB,  VPC…   h?p://aws.amazon.com/jp  
  • Amazon Cloud Terminology See http://aws.amazon.com/jp for Japanese This is not a full list of Amazon Web Service features•  AWS  –  Amazon  Web  Services  (common  name  for  Amazon  cloud)  •  AMI  –  Amazon  Machine  Image  (archived  boot  disk,  Linux,  Windows  etc.  plus  applicaJon  code)  •  EC2  –  ElasJc  Compute  Cloud   –  Range  of  virtual  machine  types  m1,  m2,  c1,  cc,  cg.  Varying  memory,  CPU  and  disk  configuraJons.   –  Instance  –  a  running  computer  system.  Ephemeral,  when  it  is  de-­‐allocated  nothing  is  kept.   –  Reserved  Instances  –  pre-­‐paid  to  reduce  cost  for  long  term  usage   –  Availability  Zone  –  datacenter  with  own  power  and  cooling  hosJng  cloud  instances   –  Region  –  group  of  Availability  Zones  –  US-­‐East,  US-­‐West,  EU-­‐Eire,  Asia-­‐Singapore,  Asia-­‐Japan  •  ASG  –  Auto  Scaling  Group  (instances  booJng  from  the  same  AMI)  •  S3  –  Simple  Storage  Service  (h?p  access)  •  EBS  –  ElasJc  Block  Storage  (network  disk  filesystem  can  be  mounted  on  an  instance)  •  RDB  –  RelaJonal  Data  Base  (managed  MySQL  master  and  slaves)  •  SDB  –  Simple  Data  Base  (hosted  h?p  based  NoSQL  data  store)  •  SQS  –  Simple  Queue  Service  (h?p  based  message  queue)  •  SNS  –  Simple  NoJficaJon  Service  (h?p  and  email  based  topics  and  messages)  •  EMR  –  ElasJc  Map  Reduce  (automaJcally  managed  Hadoop  cluster)  •  ELB  –  ElasJc  Load  Balancer  •  EIP  –  ElasJc  IP  (stable  IP  address  mapping  assigned  to  instance  or  ELB)  •  VPC  –  Virtual  Private  Cloud  (extension  of  enterprise  datacenter  network  into  cloud)  •  IAM  –  IdenJty  and  Access  Management  (fine  grain  role  based  security  keys)  
  • “The  cloud  lets  its  users  focus   on  delivering  differenAaAng   business  value  instead  of   wasAng  valuable  resources   on  the  undifferen)ated   heavy  li0ing  that  makes   up  most  of  IT   infrastructure.”      Werner  Vogels    Amazon  CTO    
  • We  want  to  use  clouds,  we  don’t  have  Jme  to  build  them   Public  cloud  for  agility  and  scale   AWS  because  they  are  big  enough  to  allocate  thousands   of  instances  per  hour  when  we  need  to  
  • Ne#lix  EC2  Instances  per  Account   (summer  2010,  producJon  is  much  higher  now…)  “Many  Thousands”   Content  Encoding   Test  and  ProducJon   Log  Analysis   “Several  Months”  
  • Ne#lix  Deployed  on  AWS  Content   Logs   Play   WWW   API   Video   S3   DRM   Search   Metadata   Masters   EMR   CDN   Movie   Device   EC2   Hadoop   rouJng   Choosing   Config   TV  Movie   S3   Hive   Bookmarks   RaJngs   Choosing   Business   Mobile   CDN   Logging   Similars   Intelligence   iPhone  
  • Cloud  Encoding  Pipeline   Encode   S3   Encode   S3  Movie   Master   Network   S3   Copy  to   CDN   Stream  Studios   Ne#lix   Master   Mezza-­‐ Mezza-­‐ to    50+   Origin   Origin   Tapes   Upload   nine   files   CDN   to  TV   nine   files   Licensed  content  is  provided  to  Ne#lix  as  high  quality  master  tapes   Many  formats  are  reduced  to  a  single  high  quality  mezzanine  format  on  S3   Individual  formats  and  speeds  are  encoded  in  over  50  combinaJons    Many  formats  for  older  and  newer  hardware  and  various  game  consoles    Many  speeds  from  mobile  through  standard  and  high  definiJon   StaJc  files  are  copied  to  each  Content  Delivery  Network’s  “origin  server”   CDNs  migrate  files  to  “edge  servers”  near  the  end  user   Files  stream  to  PC/Mac/iPad  or  TV  over  HTTP  using  “range  get”  to  move  chunks  
  • Cloud  Architecture  
  • Product  Trade-­‐off  User  Experience   ImplementaJon   Consistent   Development   Experience   complexity   OperaJonal   Low  Latency   complexity  
  • Ne#lix  Cloud  Goals  •  Faster   –  Lower  latency  than  the  equivalent  datacenter  web  pages  and  API  calls   –  Measured  as  mean  and  99th  percenJle   –  For  both  first  hit  (e.g.  home  page)  and  in-­‐session  hits  for  the  same  user  •  Scalable   –  Avoid  needing  any  more  datacenter  capacity  as  subscriber  count  increases   –  No  central  verJcally  scaled  databases   –  Leverage  AWS  elasJc  capacity  effecJvely  •  Available   –  SubstanJally  higher  robustness  and  availability  than  datacenter  services   –  Leverage  mulJple  AWS  availability  zones   –  No  scheduled  down  Jme,  no  central  database  schema  to  change  •  ProducJve   –  OpJmize  agility  of  a  large  development  team  with  automaJon  and  tools   –  Leave  behind  complex  tangled  datacenter  code  base  (~8  year  old  architecture)   –  Enforce  clean  layered  interfaces  and  re-­‐usable  components  
  • Old  Datacenter  vs.  New  Cloud  Arch   Central  SQL  Database   Distributed  Key/Value  NoSQL   SJcky  In-­‐Memory  Session   Shared  Memcached  Session   Cha?y  Protocols   Latency  Tolerant  Protocols   Tangled  Service  Interfaces   Layered  Service  Interfaces   Instrumented  Code   Instrumented  Service  Pa?erns   Fat  Complex  Objects   Lightweight  Serializable  Objects   Components  as  Jar  Files   Components  as  Services  
  • Learnings  •  Datacenter  oriented  tools  don’t  work   –  Ephemeral  instances   –  High  rate  of  change   –  Need  too  much  hand-­‐holding  and  manual  setup  •  Cloud  Tools  Don’t  Scale  for  Enterprise   –  Too  many  tools  are  “Startup”  oriented   –  Built  our  own  tools  for  1000’s  of  instances   –  Drove  vendors  to  be  dynamic,  scale,  add  APIs  •  Un-­‐modified  Datacenter  Apps  are  Fragile   –  Too  many  datacenter  oriented  assumpJons   –  We  re-­‐wrote  our  code  base!   –  (We  re-­‐write  it  conJnuously  anyway)  
  • Ne#lix  Systems  Architecture  
  • API   AWS  EC2   Front  End  Load  Balancer   Discovery   Service   API  Proxy   API  etc.   Load  Balancer   Component   API   SQS   Services   Oracl e   Oracle   Oracle   memcached   memcached   ReplicaJon   EBS   Ne?lix   S3   Data  Center  AWS  Storage   SimpleDB  
  • Database  MigraJon  •  Why  SimpleDB?   –  No  DBA’s  in  the  cloud,  Amazon  hosted  service   –  Work  started  two  years  ago,  fewer  viable  opJons   –  Worked  with  Amazon  to  speed  up  and  scale  SimpleDB  •  AlternaJves?   –  Rolling  out  Cassandra  as  “upgrade”  from  SimpleDB   –  Need  several  opJons  to  match  use  cases  well  •  Detailed  NoSQL  and  SimpleDB  Advice   –  Sid  Anand    -­‐  QConSF  Nov  5th  –  Ne#lix’  TransiJon  to  High   Availability  Storage  Systems   –  Blog  -­‐  h?p://pracJcalcloudcompuJng.com/   –  Download  Paper  PDF  -­‐  h?p://bit.ly/bhOTLu  
  • Cloud  OperaJons   Model  Driven  Architecture  Capacity  Planning  &  Monitoring  
  • Tools  and  AutomaJon  •  Developer  and  Build  Tools   –  Jira,  Perforce,  Eclipse,  Jeeves,  Ivy,  ArJfactory   –  Builds,  creates  .war  file,  .rpm,  bakes  AMI  and  launches  •  Custom  Ne#lix  ApplicaJon  Console   –  AWS  Features  at  Enterprise  Scale  (hide  the  AWS  security  keys!)   –  Auto  Scaler  Group  is  unit  of  deployment  to  producJon  •  Open  Source  +  Support   –  Apache,  Tomcat,  Cassandra,  Hadoop,  OpenJDK,  CentOS  •  Monitoring  Tools   –  Keynote  –  service  monitoring  and  alerJng   –  AppDynamics  –  Developer  focus  for  cloud  h?p://appdynamics.com   –  EpicNMS  –  flexible  data  collecJon  and  plots  h?p://epicnms.com   –  Nimso<  NMS  –  ITOps  focus  for  Datacenter  +  Cloud  alerJng  
  • Model  Driven  Architecture  •  Datacenter  PracJces   –  Lots  of  unique  hand-­‐tweaked  systems   –  Hard  to  enforce  pa?erns  •  Model  Driven  Cloud  Architecture   –  Perforce/Ivy/Jeeves  based  builds  for  everything   –  Every  producJon  instance  is  a  pre-­‐baked  AMI   –  Every  applicaJon  is  managed  by  an  Autoscaler   No  excep)ons,  every  change  is  a  new  AMI  
  • High  Availability  Zones  •  Each  zone  is  a  separate  datacenter   –  Private  power,  cooling,  network  connecJons   –  Located  close  together  for  low  latency  •  ASG  Instances  are  distributed  over  3  zones  •  Data  wri?en  to  one  zone  appears  in  all  zones  •  Ne#lix  can  survive  total  failure  of  one  zone   –  Increase  capacity  of  exisJng  zones  by  50%   –  Small  or  zero  downJme  
  • Region  MigraJon  (Ne#lix  is  working  to  have  this  in  place  during  2011,  for  internaJonal  roll-­‐out   and  disaster  recovery)  •  Data  is  backed  up  into  a  different  cloud  region   –  Cloud  bandwidth  is  much  higher  than  Datacenter  •  Restore  to  a  new  region   –  “A  few  hours”  to  load  data  and  create  databases  •  Create  model  driven  architecture   –  “A  few  hours”  to  create  service  instances  and  test  •  Send  traffic  to  new  region   –  Setup  DNS  records  and  start  customer  service  
  • Model  Driven  ImplicaJons  •  Automated  “Least  Privilege”  Security   –  Tightly  specified  security  groups   –  Fine  grain  IAM  keys  to  access  AWS  resources   –  Performance  tools  security  and  integraJon  •  Model  Driven  Performance  Monitoring   –  Hundreds  of  instances  appear  in  a  few  minutes…   –  Tools  have  to  “garbage  collect”  dead  instances    
  • Ne#lix  App  Console  
  • Auto  Scale  Group  ConfiguraJon  
  • Capacity  Planning  &  Monitoring  
  • Capacity  Planning  in  Clouds   (a  few  things  have  changed…)  •  Capacity  is  expensive  •  Capacity  takes  Jme  to  buy  and  provision  •  Capacity  only  increases,  can’t  be  shrunk  easily  •  Capacity  comes  in  big  chunks,  paid  up  front  •  Planning  errors  can  cause  big  problems  •  Systems  are  clearly  defined  assets  •  Systems  can  be  instrumented  in  detail  •  Depreciate  assets  over  3  years  (reservaJons!)  
  • Monitoring  Issues  •  Problem   –  Too  many  tools,  each  with  a  good  reason  to  exist   –  Hard  to  get  an  integrated  view  of  a  problem   –  Too  much  manual  work  building  dashboards   –  Tools  are  not  discoverable,  views  are  not  filtered  •  SoluJon   –  Get  vendors  to  add  deep  linking  URLs  and  APIs   –  IntegraJon  “portal”  Jes  everything  together   –  Underlying  dependency  database   –  Dynamic  portal  generaJon,  relevant  data,  all  tools  
  • Data  Sources   • External  URL  availability  and  latency  alerts  and  reports  –  Keynote   External  TesJng   • Stress  tesJng  -­‐  SOASTA   • Ne#lix  REST  calls  –  Chukwa  to  DataOven  with  GUID  transacJon  idenJfier   Request  Trace  Logging   • Generic  HTTP  –  AppDynamics  service  Jer  aggregaJon,  end  to  end  tracking   • Tracers  and  counters  –  log4j,  tracer  central,  Chukwa  to  DataOven   ApplicaJon  logging   • Trackid  and  Audit/Debug  logging  –  DataOven,  Appdynamics    GUID  cross  reference   • ApplicaJon  specific  real  Jme  –  Nimso<,  Appdynamics,  Epic   JMX    Metrics   • Service  and  SLA  percenJles  –  Nimso<,  Appdynamics,  Epic,logged  to  DataOven   • Stdout  logs  –  S3  –  DataOven,  Nimso<  alerJng  Tomcat  and  Apache  logs   • Standard  format  Access  and  Error  logs  –  S3  –  DataOven,  Nimso<  AlerJng   • Garbage  CollecJon  –  Nimso<,  Appdynamics   JVM   • Memory  usage,  call  stacks,  resource/call  -­‐  AppDynamics   • system  CPU/Net/RAM/Disk  metrics  –  AppDynamics,  Epic,  Nimso<  AlerJng   Linux   • SNMP  metrics  –  Epic,  Network  flows  -­‐  FasJp   • Load  balancer  traffic  –  Amazon  Cloudwatch,  SimpleDB  usage  stats   AWS   • System  configuraJon    -­‐  CPU  count/speed  and  RAM  size,  overall  usage  -­‐  AWS  
  • Integrated  Dashboards  
  • Dashboards  Architecture  •  Integrated  Dashboard  View   –  Single  web  page  containing  content  from  many  tools   –  Filtered  to  highlight  most  “interesJng”  data  •  Relevance  Controller   –  Drill  in,  add  and  remove  content  interacJvely   –  Given  an  applicaJon,  alert  or  problem  area,  dynamically   build  a  dashboard  relevant  to  your  role  and  needs  •  Dependency  and  Incident  Model   –  Model  Driven  -­‐  Interrogates  tools  and  AWS  APIs   –  Document  store  to  capture  dependency  tree  and  states  
  • Dashboard  Prototype   (not  everything  is  integrated  yet)  
  • AppDynamics   How  to  look  deep  inside  your  cloud  applicaJons  •  AutomaJc  Monitoring   –  Base  AMI  bakes  in  all  monitoring  tools   –  Outbound  calls  only  –  no  discovery/polling  issues   –  InacJve  instances  removed  a<er  a  few  days    •  Incident  Alarms  (deviaJon  from  baseline)   –  Business  TransacJon  latency  and  error  rate   –  Alarm  thresholds  discover  their  own  baseline   –  Email  contains  URL  to  Incident  Workbench  UI  
  • Using  AppDynamics  (simple  example  from  early  2010)  
  • Point  Finger  and  Assess  Impact   (an  async  S3  write  was  slow,  no  big  deal)  
  • Monitoring  Summary  •  Broken  datacenter  oriented  tools  is  a  big  problem  •  IntegraJng  many  different  tools   –  They  are  not  designed  to  be  integrated   –  We  have  “persuaded”  vendors  to  add  APIs  •  If  you  can’t  see  deep  inside  your  app,  you’re  L  
  • Wrap  Up  
  • ImplicaJons  for  IT  OperaJons  •  Cloud  is  run  by  developer  organizaJon   –  Our  IT  department  is  Amazon  Cloud  •  Cloud  capacity  is  much  bigger  than  Datacenter   –  Datacenter  oriented  IT  staffing  is  flat   –  We  have  no  IT  staff  working  on  cloud   –  We  have  moved  3  people  out  of  IT  to  write  code  •  TradiJonal  IT  Roles  are  going  away   –  Don’t  need  SA,  DBA,  Storage,  Network  admins  
  • Next  Few  Years…  •  “System  of  Record”  moves  to  Cloud  (now)   –  Master  copies  of  data  live  only  in  the  cloud,  with  backups   –  Cut  the  datacenter  to  cloud  replicaJon  link  •  InternaJonal  Expansion  –  Global  Clouds  (later  in  2011)   –  Rapid  deployments  to  new  markets  •  Cloud  StandardizaJon?   –  Cloud  features  and  APIs  should  be  a  commodity  not  a  differenJator   –  DifferenJate  on  scale  and  quality  of  service   –  CompeJJon  also  drives  cost  down   –  Higher  resilience  and  scalability     We  would  prefer  to  be  an  insignificant  customer  in  a  giant  cloud  
  • Takeaway    Ne9lix  is  path-­‐finding  the  use  of  public  AWS   cloud  to  replace  in-­‐house  IT  for  non-­‐trivial  applicaAons  with  hundreds  of  developers  and   thousands  of  systems.     acockcro<@ne#lix.com   h?p://www.linkedin.com/in/adriancockcro<   @adrianco  #ne#lixcloud