EMC World 2008: Documentum 6.5 Architecture Overview

Time for an update on the architecture. Victor Spivak is presenting. He knows his stuff, so hopefully it’ll be a good one. I know waaay too many people in this room. I’ll never get to say hello to all of them. This is by far the most crowded session.

• He had two presentation slots and the crowd voted for a more detailed two-part presentation rather than one high-level and one deep-dive, neither covering everything.
• Layered Technical Architecture (bottom-up)
  • Repository Infrastructure (database, index…)
  • Repository Services
  • Security
  • Compliance
  • Content Services
  • Process Services
  • Client Infrastructure Layer
• Tools, Programming Interfaces, and Configuration Interfaces across all but the Infrastructure Layer
• D6.5 adds XML Store on the Infrastructure Layer (formerly X-Hive)
  • Allows for rapid XQuery against the XML in the system
• SOA is a focus on where they are trying to move Documentum from client-server, New Paradigm
  • Easy to do from scratch, hard to transition
  • Started with 5.x with Business Office Framework
  • DFS in D6
• DFS has always been JAX-WS based in terms of its internal runtime foundation. The previous Web Services Foundation (WSF), which just project SBO’s as web services but didn’t provide out-of-the-box (OOTB) services like DFS does, was Axis-based. (From Craig Randall per comment below).
  
• They provide client-side libraries to simplify development in Java and .NET (D6.5 sp1)
  • This is to aid development.
  • Can still use WSDL right now.
• Services (first 6 are in D6):
  • Schema: Data Dictionary
  • Object: Create, Update, Delete, Copy Move, Validate…etc.
  • Version: checkin, checkout, cancelCheckout, etc.
  • Query: DQL (supports paging)
  • Search: support mulltiple repositories, can tie into ECIS
  • Workflow: startProcess, getProcessInfo, etc.
  • Access Control
  • Lifecycle
  • Virtual Documents
  • Query Store: Saved searches
  • Task Management
  • Analytics
  • Content Delivery
  • Profile (CTS)
  • Transformation
  • Policy: apply, remove, getAppliedPolicies, etc. (Retention Policy Services)
  • Electronic Signature: add, verify
  • ERP Integration: execute Action, executeAgent, etc.
  • Formal Record
• Three types of services in the Intelligent Information Infrastructure: Content, Archive, and XML
• Post D6.5
  • More services
  • Services governing strict process (ensure services ineroperability)
  • Support WS Standards
  • REST support: Define resource model and URI (complimentary to the current SOAP-based services)
  • Next generation of clients will use SOA
• May preclude DFC from the application layer in the future.
• As last year, Aspects are presented as the way to slim the system object, and thus everything we throw into Documentum
• Operation Customization D6.5 (already in there, just not documented)
  • Able to modify the behavior of action on the server-side without modifying the application
  • Import, Export, Checkout, Check-in, Cancel Checkout, Delete, Copy, Move, Validate, Transform
  • Defines steps for each operation
  • Allows operation cutomization based on Type, Format, or Repository
  • Includes modifying existing operations and creating new operations
  • Register Operation populators
  • Can add new steps and replace existing actions
  • Using BOF 2
• BOF 3 in post D6.5
  • BOF versioning, allows different versions of BOF objects to co-exist
  • Will add notion of bundling of business objects, tracking those dependencies
  • Tighter integration with Composer (I hope so because it is miserable now)
  • Simplified customization model
• Lightweight sysobject
  • Current sysobject 3K+ per instance (I always thought it was 2K)
  • Light ones allow multiple instances to share the same set of policies and metadata
  • One instance of Parent object and one smaller child object for each individual object
  • Changing the Parent impacts all children
  • Make use of Documentum as a Archive repository much more realistic
  • Will complement existing object model
  • Object look and behave like normal sysobject
  • Objects can be change to heavyweight sysobject from lightweight sysobject if needed so shared values on Parent can be changed without impacting the other children.
  • 30x footprint reduction seen in database, depends on number of parents to children
• Batching
  • Allows the grouping of multiple operations into one set of SQL statements to the database
  • IDfBatchManager interface allows you to open and then close a batch. normal DFC in the middle
  • Helps with mass import, reduces round-trips to database
• Scoping
  • Server performs checks during every operation (does this folder exist)
  • Allows for suspension of checks when performing a batch operaion
  • Default Scopes: Operation, Transaction, Batch
  • When combined with Batching, getting 3x+ performance improvement
• Partitioning
  • Allows items to be placed in logical partition to manage items and to store.
• JBoss as application server in D6.5, option or replacement? Need to ask Product Manager
• Indexing Architecture in D7
  • Lucene (Had been D6.5, very sad)
  • support for XQuery

Off to the Standards session.

Disclaimer

All information in this post was gathered from the presenters and presentation. It does not reflect my opinion unless clearly indicated (Italics in parenthesis). Any errors are most likely from my misunderstanding a statement or imperfectly recording the information. Updates to correct information are reflected in red, but will not be otherwise indicated.

All statements about the future of EMC products and strategy are subject to change due to a large variety of factors.