Alfresco2CAStor

1
Alfresco2CAStor

• Bridging Alfresco and CAStor

R. Timmermans, Partner _at_ XeniT
2
A New Alfreso file storebusiness case Optimal
solution for 10 million documents, guaranteed
immutable, safely stored, with redundancy, free
of backup insurance archive
3
What is CAStor?
Information Lifecycle Management (ILM) Software
on top of linux to run a storage cluster,
bootable from USB stick
Traditionally complex
Innovatively simple
Applications
Applications
Mgmt.
SAN
NAS
Dynamic SAN, NAS
Tape Archive
pumps
90 fixed content CAStor
4
The content matters, not the location

• Basic Concept of Content Addressed Storage is
  Simplicity
• Put file in
• Get unique identifier key back
• Save key in doc or database
• Provide key to get file back
• Additional functionality
• Define file retention policies
• All the way from read/write to WORM
• Define number of file replicas
• Include user metadata with each file
• Universal HTTP interface any environment can use

5
CAStor automatically balances migrates!
Add new hardware
Retire old hardware
Hardware Life Cycle Management
6
CAStor Environment
CAStor Cluster
7
XeniT Caringo partnership

• XeniT focused Alfresco integrator, we do only
  Alfresco
• Caringo content storage simplified
• Alfresco CAStor Interface
• The Alfresco CAStor Interface, developed by
  XeniT, is available as open source at
  http//forge.alfresco.com/projects/alfresco2castor
  /
• In its initial implementation, the Alfresco
  CAStor Interface supports the usage of CAStor as
  an Alfresco content repository and an archive for
  storing and retaining regulated and immutable
  documents/content over the long-term.
• XeniT offers
• installation, configuration and certification
  services
• a maintenance contract on a supported release
• new features in the supported version

8
CastorContentStore in Alfreso

• 100 HTTP based (Restful)
• Content URL reference in the format
  'castor//castorClusterName/UUID'.
• Converted via contentUrl2CastorUrl
• castor//clusterName/UUIDmimetype.... gt
  http//castorNode/UUID
• CastorContentStore
• Alternative store implementation alf-castor.jar
• in ALFRESCO_HOME/tomcat/webapps/alfresco/WEB-INF/l
  ib
• add the CastorContentStore as a bean
• ALFRESCO_HOME/tomcat/shared/classes/alfresco/exten
  sion/custom-repository-context.xml
• ltbean id"castorContentStore"
  class"eu.xenit.alfresco.castor.CastorContentStore
  "gt
• lt!-- clusterName --gt ltconstructor-arg index"0"
  value"testCluster"/gt
• lt!-- list of clusterNodes --gt
• ltconstructor-arg index"1"gt ltlistgt
  ltvaluegt10.0.1.1lt/valuegt ltvaluegt10.0.1.2lt/valuegt
  ltvaluegt10.0.1.3lt/valuegt lt/listgt
  lt/constructor-arggt
• lt!-- maxTotalNbrOfConnections --gt
  ltconstructor-arg index"2" value"50"/gt
• lt!-- maxNbrOfConnectionsPerHost --gt
  ltconstructor-arg index"3" value"20"/gt
• lt!-- timeout (used for socketTimeout
  1connectionTimeout in milliseconds)
  -ltconstructor-arg index"4" value"5000"/gt
• lt!-- maxNbrOfClusterAccessAttempts --gt
  ltconstructor-arg index"5" value"3"/gt
• lt!-- lifePoint parameter --gt ltconstructor-arg
  index"6" value" minreps2, maxreps3,
  deletableno"/gt
• CastorContentStore content store object, which
  maintains connections to CAStor servers in the
  cluster.

9
CastorContentStore part II

• configure Alfresco to use the new store
• fileContentStore is referenced in several places
  
• avm-services-context.xml avmRepository
• content-services-context.xml contentStoreCleaner
  
• content-services-context.xml contentService
• core-services-context.xml avmLuceneIndexerAndSear
  cherFactory
• sufficient to override the
• contentStoreCleaner and
• contentService beans
• with instances that refer to 'castorContentStore'
  (instead of 'fileContentStore').
• Note We haven't used CAStor in combination with
  avm.

10
(No Transcript)
11
(No Transcript)
12
Demo

• Connect with VNC
• Open castor monitor screen
• Add document
• Open document
• View log with CAStor url
• Check immediate CAStor url
• Remove connection
• Open document (retrieved from other node)

13
Alfresco2Castor Status

• Production quality to store and retrieve files
  (5 million docs at Vivium insurance)
• Added thread-safeness to alfresco2castor in V1.1
• Combined with fast, multi-threaded input programs
  to load archives at 1000 documents per minute
• Includes creating subspace, setting access
  control, meta-data, ...
• CAStor is NOT the bottleneck!
• Content router Replication between
  geographical sites (Antwerp Brussels)
• Roadmap
• Integrated clean-up delete
• Efficiënt redirect for distributed Alfresco
• Seamless store switch
• integrating code from P. Dubois
• Aspect or rule based

14
Alfresco quote

• The idea of connecting an ECM to a virtually
  infinitely scalable storage cloud is something
  Im very excited about and finding that XeniT, an
  Alfresco and Caringo partner, have published the
  source code to the Alfresco Forge pleases me to
  no end.
• Luis Sala, Senior Director of Solutions
  Engineering Alfresco Software, Inc.

15
For more information

• www.xenit.eu
• info_at_xenit.eu
• sales_at_xenit.eu