Inside Documentum

Momentum 2011 pt 3 Jerome’s Architecture Session

jeroem has another session tomorrow evening, The Innovation Engine, so a lot of the really cool new stuff will be. Shown then. However this is what he did show this morning. As usual we overran-EMC how About making this slot longer ?

6.7 loads of perf improvements, some examples (too many on the slide to copy down):
64bit Content Server 4:1 reduction of Content server instances
Input accel pages/day 500k, tasks/day 2m
Doc science improvements
Bam 10k reports/day consume 15 m events/day

Current stack has innovation everywhere-much if it arising from future work on NgiS

High Avail:
Bocs becoming smarter in predictive caching

Deploy efficiency.
Xms:
blueprint defines requirement ,
Blueprints are environment independent
Each vm has hyperic agents. Centralised hyperic db (telemetry project). Dashboard hyperic client.
Adjustment , ie reaction based on hyperic data

Integrion??? Feed it with hyperic data and learns behaviour. Lead to auto remedy when application shows abnormal behaviour

Session pooling. Analysed context switch. Look to have more linear session increments – better scalability (nb dctm already really good at this)

Showed metric
10k getsessions 6min -> 8 sec
10k getcontent 7.5 min -> 34sec

New services: Ajax-> rest services. Use cases
1) content services incl cmis
2) generated Service eg invoice object type incl generated restful service

“xcp misunderstood “, app modelling (semantic model ) then generate runtime. Ultimately leads to auto deployment. Don’t just focus on ui.

D7 dormant state feature
Rolling upgrade + others stuff I didn’t capture

Ngis:
Embed greenplum
Impl xacml

“big information” “bring processing to the data”
Realtime map reduce (traditional map reduce too slow)

Kazeon crawlers!