Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Michael Nitschinger on Couchbase

Michael Nitschinger on Couchbase

More Decks by Enterprise Java User Group Austria

Other Decks in Technology

Transcript

  1. Who‘s  this  guy?   •  SDK  Engineer  at  Couchbase  

    ­  Maintaining   •  Official  Java  SDK   •  Spymemcached   •  Spring-­‐Data-­‐Couchbase   •  @daschl  on  TwiCer   •  Core  Team  of  NeCy.io   •  ContribuGng  to  Reactor  and  others    
  2. Instagram gained nearly 1 million users overnight when then expanded

    to Android Growth  is  the  New  Reality  
  3. Draw  Something  Viral  Growth   19   17   15

      13   11   9   7   5   3   3/1   28   26   24   22   20   18   16   14   12   10   8   2/6   Draw  Something  by  OMGPOP   Daily  Ac)ve  Users  (millions)   21   2   4   6   8   10   12   14   16  
  4. How  do  you  take  this  growth?   RDBMS  is  good

     for  many  thing,  but  hard  to  scale   RDBMS  Scales  Up   Get  a  bigger,  more  complex  server   Users   ApplicaUon  Scales  Out   Just  add  more  commodity  web  servers   Users   System  Cost   ApplicaUon  Performance     RelaUonal  Database   Web/App  Server  Tier   System  Cost   ApplicaUon  Performance     Won’t  scale   beyond  this   point  
  5. Scaling  out  RDBMS   • Run  Many  SQL  Servers   • Data

     could  be  shared   -­‐ Done  by  the  applicaUon  code   • Caching  for  faster  response  Ume   Web/App  Server  Tier   Memcached  Tier   MySQL  Tier  
  6. Scaling  out  fla]ens  the  cost  and  performance  curves   NoSQL

     Database  Scales  Out   Cost  and  performance  mirrors  app  Uer   Users   NoSQL  Distributed  Data  Store   Web/App  Server  Tier   ApplicaUon  Scales  Out   Just  add  more  commodity  web  servers   Users   System  Cost   ApplicaUon  Performance     ApplicaUon  Performance     System  Cost   NoSQL  Technology  Scales  Out  
  7. Couchbase  Open  Source  Project   •  Leading  NoSQL  database  project

      focused  on  distributed  database   technology  and  surrounding   ecosystem   •  Supports  both  key-­‐value  and   document-­‐oriented  use  cases   •  All  components  are  available   under  the  Apache  2.0  Public   License   •  Obtained  as  packaged  so`ware  in   both  enterprise  and  community   ediUons.  
  8. Easy   Scalability   Consistent  High   Performance   Always

     On   24x365   Grow  cluster  without   applicaUon  changes,  without   downUme  with  a  single  click   Consistent  sub-­‐millisecond     read  and  write  response  Umes     with  consistent  high  throughput   No  downUme  for  so`ware   upgrades,  hardware   maintenance,  etc.   Flexible  Data   Model   JSON  document  model  with  no   fixed  schema.   Couchbase  Server  Core  Principles  
  9. New  in  2.0+   JSON  support   Indexing  and  Querying

      Cross  data  center  replicaGon   Incremental  Map  Reduce   JSON JSON JSON JSON JSON
  10. Couchbase  Server  Architecture   Heartbeat   Process  monitor   Global

     singleton  supervisor   ConfiguraUon  manager   on  each  node   Rebalance  orchestrator   Node  health  monitor   one  per  cluster   vBucket  state  and  replicaUon  manager   hCp   REST  management  API/Web  UI   HTTP   8091   Erlang  port  mapper   4369   Distributed  Erlang   21100  -­‐  21199   Erlang/OTP   storage  interface   Couchbase  EP  Engine   11210   Memcapable    2.0   Moxi   11211   Memcapable    1.0   Memcached   New  Persistence  Layer   8092   Query  API   Query  Engine   Data  Manager   Cluster  Manager  
  11. Couchbase  Server  Architecture   ReplicaUon,  Rebalance,     Shard  State

     Manager   REST  management     API/Web  UI   8091   Admin  Console   Erlang  /OTP   11210  /  11211   Data  access  ports   Object-­‐managed   Cache   MulG-­‐threaded     Persistence  Engine   8092   Query  API   Query  Engine   hCp   Data  Manager   Cluster  Manager  
  12. 3   3   2   Single  node  -­‐  Couchbase

     Write  OperaGon   Managed  Cache   Disk  Queue   Disk   ReplicaUon   Queue   App  Server   Couchbase  Server  Node   Doc  1   Doc  1   Doc  1   To  other  node  
  13. COUCHBASE  SERVER    CLUSTER   Cluster-­‐wide  Basic  OperaGon   • 

    Docs  distributed  evenly  across   servers     •  Each  server  stores  both  acGve  and   replica  docs   Only  one  server  acUve  at  a  Ume   •  Client  library  provides  app  with   simple  interface  to  database   •  Cluster  map  provides  map     to  which  server  doc  is  on   App  never  needs  to  know   •  App  reads,  writes,  updates  docs   •  MulGple  app  servers  can  access  same   document  at  same  Gme   User  Configured  Replica  Count  =  1   READ/WRITE/UPDATE       ACTIVE   Doc  5   Doc  2   Doc   Doc   Doc   SERVER  1       ACTIVE   Doc  4   Doc  7   Doc   Doc   Doc   SERVER  2   Doc  8       ACTIVE   Doc  1   Doc  2   Doc   Doc   Doc   REPLICA   Doc  4   Doc  1   Doc  8   Doc   Doc   Doc   REPLICA   Doc  6   Doc  3   Doc  2   Doc   Doc   Doc   REPLICA   Doc  7   Doc  9   Doc  5   Doc   Doc   Doc   SERVER  3   Doc  6   APP  SERVER  1   COUCHBASE  Client  Library       CLUSTER  MAP   COUCHBASE  Client  Library       CLUSTER  MAP   APP  SERVER  2   Doc  9  
  14. Add  Nodes  to  Cluster   •  Two  servers  added  

    One-­‐click  operaGon   •  Docs  automaGcally   rebalanced  across   cluster   Even  distribuUon  of  docs   Minimum  doc  movement   •  Cluster  map  updated   •  App  database     calls  now  distributed     over  larger  number  of   servers         REPLICA   ACTIVE   Doc  5   Doc  2   Doc   Doc   Doc  4   Doc  1   Doc   Doc   SERVER  1       REPLICA   ACTIVE   Doc  4   Doc  7   Doc   Doc   Doc  6   Doc  3   Doc   Doc   SERVER  2       REPLICA   ACTIVE   Doc  1   Doc  2   Doc   Doc   Doc  7   Doc  9   Doc   Doc   SERVER  3       SERVER  4       SERVER  5   REPLICA   ACTIVE   REPLICA   ACTIVE   Doc   Doc  8   Doc   Doc  9   Doc   Doc  2   Doc   Doc  8   Doc   Doc  5   Doc   Doc  6   READ/WRITE/UPDATE   READ/WRITE/UPDATE   APP  SERVER  1   COUCHBASE  Client  Library       CLUSTER  MAP   COUCHBASE  Client  Library       CLUSTER  MAP   APP  SERVER  2   COUCHBASE  SERVER    CLUSTER   User  Configured  Replica  Count  =  1  
  15. Fail  Over  Node       REPLICA   ACTIVE  

    Doc  5   Doc  2   Doc   Doc   Doc  4   Doc  1   Doc   Doc   SERVER  1       REPLICA   ACTIVE   Doc  4   Doc  7   Doc   Doc   Doc  6   Doc  3   Doc   Doc   SERVER  2       REPLICA   ACTIVE   Doc  1   Doc  2   Doc   Doc   Doc  7   Doc  9   Doc   Doc   SERVER  3       SERVER  4       SERVER  5   REPLICA   ACTIVE   REPLICA   ACTIVE   Doc  9   Doc  8   Doc   Doc  6   Doc   Doc   Doc  5   Doc   Doc  2   Doc  8   Doc   Doc   •  App  servers  accessing  docs   •  Requests  to  Server  3  fail   •  Cluster  detects  server  failed   Promotes  replicas  of  docs  to   acUve   Updates  cluster  map   •  Requests  for  docs  now  go  to   appropriate  server   •  Typically  rebalance     would  follow   Doc   Doc  1   Doc  3   APP  SERVER  1   COUCHBASE  Client  Library       CLUSTER  MAP   COUCHBASE  Client  Library       CLUSTER  MAP   APP  SERVER  2   User  Configured  Replica  Count  =  1   COUCHBASE  SERVER    CLUSTER  
  16. COUCHBASE  SERVER    CLUSTER   Indexing  and  Querying    

    User  Configured  Replica  Count  =  1       ACTIVE   Doc  5   Doc  2   Doc   Doc   Doc   SERVER  1   REPLICA   Doc  4   Doc  1   Doc  8   Doc   Doc   Doc   APP  SERVER  1   COUCHBASE  Client  Library       CLUSTER  MAP   COUCHBASE  Client  Library       CLUSTER  MAP   APP  SERVER  2   Doc  9   •  Indexing  work  is  distributed   amongst  nodes   •  Large  data  set  possible   •  Parallelize  the  effort   •  Each  node  has  index  for  data  stored   on  it   •  Queries  combine  the  results  from   required  nodes       ACTIVE   Doc  5   Doc  2   Doc   Doc   Doc   SERVER  2   REPLICA   Doc  4   Doc  1   Doc  8   Doc   Doc   Doc   Doc  9       ACTIVE   Doc  5   Doc  2   Doc   Doc   Doc   SERVER  3   REPLICA   Doc  4   Doc  1   Doc  8   Doc   Doc   Doc   Doc  9   Query  
  17.     SERVER  3       SERVER  1  

        SERVER  2   Couchbase  Server  –  San  Francisco       SERVER  3       SERVER  1       SERVER  2   Couchbase  Server  –  New  York   OpGmisGc  replicaGon   Per  replicaGon     Tunable  Parameters   OpGmized  protocol   based  on  memcached   Reliability  and     performance  at  scale   Cross  Data  Center  ReplicaGon  (XDCR)  
  18. Official  Java  SDK   •  Current  Release:  1.2.3  (December  2013)

      •  Fully  asynchronous  (Futures  &  Callbacks)   •  Builds  on  the  (very  mature)  Spymemcached  Library   •  Does  the  heavy  liping  for  you   ­  Knows  the  cluster  topology   ­  „Routes“  operaUons  to  target  servers   ­  Abstracts  protocol  semanUcs  (memcache,  h]p,...)    
  19. Spring  IntegraGon   •  IntegraGon  with  Spring  Data   ­ 

    Templates   ­  Repositories   ­  ExcepUon  Mapping     •  Transparent  Caching  support  (@Cacheable)   •  Maps  POJO  EnGGes  to  JSON  (and  back)   •  Milestone  2  released,  next  up:  RC1   •  Planned:  Spring  Boot,  Spring  XD  
  20. And  more...   •  Play  Framework  IntegraGon   ­  Community

     driven   ­  h]ps://github.com/mathieuancelin/play2-­‐couchbase   •  Working  on:   ­  Hibernate  OGM  support   ­  N1QL  Query  IntegraUon   ­  Next  GeneraUon  SDKs  (2.0  API  and  performance)