Consolidation

[resplin]

Obsolete Pages{{Obsolete}}

The official documentation is at: http://docs.alfresco.com

Design DiscussionsResearch For Future Features

Table of Contents

• 1 Consolidation Tasks


• 2 Consolidation Areas


• 3 Meeting Notes
  
  
  
  • 3.1 Session (Associations)
  
  
  • 3.2 Session (Policies)
  
  
  • 3.3 Session (CICO)
  
  
  • 3.4 Session (Admin)
  
  
  • 3.5 Session (Performance)
  
  
  • 3.6 Session
  
  

Consolidation Tasks

• Consolidation goals
  
• Review AVM in light of experience
  
• DM CICO paradigm
  
• API impact
  

Consolidation Areas

• Optimize out cache dependency
  
• Native cache management
  
• Data Dictionary
  
• Full Type support
  
• Association support
  
• Policies
  
• Export/Import (done?)
  
• Query Language
  

Meeting Notes

Session (Associations)

• Fold it down to simple child-assoc of any type
  
• Translations; Rules - associate by location or version instance
  
• Hierarchies (permissions & rules become first-class objects)
  

Session (Policies)

• Property changed
  
• Indexing @ node
  
• Constraints @ node
  
• Audit should be pushed down
  
• Cache resetting based on type
  
• Minimise 'event'
  
• Standard behaviours pushed down
  
• No pre-get
  
• Listener-based; optimization of rules
  
• Simplified set of policies
  
• Only post-policies
  
• Abstract implementation of service (async action queue)
  
• Make object transaction single (outstanding processing)
  
• Use staging concepts to scope business transactions
  
• Queue management (remove dupes; groom queue)
  
• Records management?
  
• Optional virtual 'lock'
  

Session (CICO)

• Each user has own view
  
• Skeletal projection of My Modified
  

Session (Admin)

• Node edit/remove
  
• Purge by type
  
• Webapp redeploy
  
• Back-up & restore
  
• Build (problems with circular dependencies, standard jars)
  
• JMS (add notify by JMS action; queue tasks/actions for object; possibly remote)
  
• Permissions (roles; merge app personalities & rights on object)
  
• LDAP (sync not import; delta)
  

Session (Performance)

• Zero transactionality
  
• Global indexing (e.g. Nutch/Solr)
  
• Workflow
  
• Proxy to many instances
  
• Group ids stored with user in DB
  
• Locking in cluster for CIFS (opportunistic locking)
  

Session

• Core repo with very small transactions
  
• Loose associations (relationships)
  
• Store 'transienrt' data in Hadoop-like file stores (posts, ratings, tags)
  
• Calculated views of aggregated date (events, activities, tag clouds)
  
• Design for pluggability of cluster nodes (ease of support, installation)