EMC World 2012: Documentum Architecture
Here we, as in the whole gang, are at Jeroen van Rotterdam’s session on Documentum Architecture. I have to see this, even if I never implement Documentum again. I am just too curious to not attend, plus Jeroen is a very entertaining speaker.
- Architecture looks same as before, though likely a new diagram. D2 Configuration added but nothing surprising at this point. Very high-level model.
- Trying to move away from coding content applications to either;
- Composing applications in xCP
- Configuring interfaces in D2
- Contrasting Today’s architecture with Future (NGIS-Next Generation Information Server)
- NGIS is multi-tenant Public Cloud Architecture
- Marginal cost to add a tenant needs to be almost zero
- SOAP is gone, only REST
- D2 Config part of core configuration component
- Created Documentum Markup Language to translate current model into XML to facilitate transformation from current stack to NGIS. Starting with D2, xCP next
- Goal is design against current stack and allow for flowing it between the current stack on-premise to future cloud stack off premise. Includes solution and information (likely assumes an up-to-date xCP or D2 solution)
- Setting-up NGIS authentication with either internal LDAP or OpenID.
- Investing in both the current platform and NGIS
- For D7, working on performance and scalability
- Improved Context Switch performance (reuse of session by other users)
- Improved cache management, lower memory and CPU
- Improved capacity of session and response times, reduced sessions in Oracle
- Session to User ratio is dramatically better. less than 20 sessions for 400 users. D6.7 needs 80 sessions for 100 sessions.
- Around 300 users, response times die around 100 concurrent users and spikes up. D7 session response time more than 300% better even when scaled to 500 users
- More graphs (all 64-bit), very impressive, D7 is reletively flat to 400+ users while D6.7 is spiking up at 100 users. (Even taking values aside, the ratio of performance improvement between the two is dramatic)
- This is all done through various means. Jeroen is showing a graph that just focuses on the new type cache methodology
- Going for a pure REST approach, takes time to do right
- Abstracting resource types from Documentum platform
- Supporting resource mobility (HTTP redirects)
- Connecting resources across products in seamless manner to application
- Discoverability of resources
- Unstructured Information Management Architecture (UIMA)
- Apache Standard Architecture for Natural Language Processing customizations
- Coming Released in xPlore 1.2 (Nov 2011)
- Example is Content Intelligence Services (CIS)
- Federated Search in Mobile, Webtop, and CenterStage. D2 and xCP not currently surfaced.
- (Note to all Federated Search is the reaching out to other search indexes to display one result set. Is not Enterprise Search which is one search engine and multiple source systems)
- Hyperic based monitoring solution per customer in D6.7 sp2 and onwards (later this year). Plan to have it eventually allow you to correct errors or have it automatically correct errors. (this and many tech improvements are stemming from On-Demand efforts)
- Includes Logging
- Windows and Linux
- Later this year for D6.7 sp2 is likely Q4 but no commitment for earlier. “Later this year” is catch phrase. Will have to ask others.
- XMS is designed to help automate the deployment of Documentum.
- D2 configuration based upon user and object type and a priority order
- Semantic model of xCP generates REST services
- Stateless BPM engine will allow user to trigger a process and have result appear in the interface
- xCP and Big Data
- in xCP 2.0, built-in integration for Greenplum from xCP through xCP Designer, same for Document Sciences xPression
- Stateless engine will interact to interact with Greenplum directly
- Conditional output from Greenplum can create letter from Document Sciences without user interaction, all through engine
Time for more coffee. Going to find a good spot to see Tucci’s keynote. Enjoying the conference so far, but lots to learn.
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 may not be otherwise indicated.
All statements about the future of EMC products and strategy are subject to change at any time due to a large variety of factors.