[OOTB-hive] Workdesk, Case Management?

[Heiko Robert]

Hi David,

case management and the Alfresco repository are not the best friends if 
you don't implement a separate (preferable relational) layer for the 
case and business object master data. That may be the reason why this 
road hasn't been continued.

Alfresco made some architectural decisions which works good for 
archiving (dead documents) but which conflict with the requirement to 
_relate_ objects by metadata and processes as you may have already seen 
in other case management systems:
* by default metadata is queried from SOLR (which is not designed to be 
a replacement for a relational db)
* you can't garantuee when metadata is stored in the index (async)
* alfresco search doesn't support joins / complex queries, metadata 
inheritance (I mean inheritance from data, not from model) which are 
important principles in cases
* the alfresco db schema is not optimized for mass metadata queries / 
case like queries and therfore not an alternative if you want to support 
millions of nodes except you create your own BI like autogenerated 
schema on top using mechanisms like materialized views.

With this in mind any case like layer would be very limited on vanilla 
Alfresco because you can only query on metadata directly stored on the 
node but your users want to define complex queries for their case logic, 
dimensions and they expect joined acces to other internal or external data.

We decided to go the american way: love it or leave it (as it is). But 
don't try to change it.

To be happy with alfresco as it is we only store in alfresco:
* metadata which will never change
* references to master data as link to the case objects managed / stored 
outside Alfresco (e.g. case-id, business-partner-id, process-id, ...)
Additionally we extended Share to use the external stored metadata and 
query capabilities from our mastadata and case middle tier using 
configurable controls which no longer require expensive Share 
customization to support case views and showing relevant external data.
To work around the query restrictions in Alfresco to enable expected 
case views we don't use smart folders but create associations controlled 
by our custom rule based configurable layer (you may call this links to 
other nodes). This scales much better and gives more flexibility than 
just being restricted on metadata stored on a node.

To make the long story short:
Alfresco is not the perfect layer for case management. You may want a 
separate layer for that and either another client instead of Share or 
configurable extensions for Share (as a supported, managed product) to 
implement the expected behavior and UI.
I would not encourage anybody to customize Share for this as a single 
customer project. You would be surprised about the huge efforts to 
support and migrate this over several Alfresco Share versions.

If there is enough interest from others we may discuss this in detail 
and by examples in a private web meeting but be warned: our solutions 
are not open source and not for free ;-)

Regards
Heiko

Am 21.02.2017 um 01:18 schrieb L. David Zuluaga Botero:
> Hi Bees, sorry for writing so many times, but we're pretty new to the
> community and would like to contact people who have the same questions
> we have!
>
> I am writing now because according to some client's requests and some
> inquiries I've made, seems that the perfect solution is Alfresco
> Workdesk!. (case management and similar applications)
>
> However, the project seems to be deprecated, which I cannot
> understand... Since it seems to be an increasing need from the public,
> and are integrated systems that alike systems like Nuxeo has integrated,
> according to what can be seen really quick on their web.
>
> Anyhow, I though the Alfresco Share would have something similar.. but
> there's no trace of it.
>
> Luis Sanchez, a OOTB member, suggested us to retake the most recent
> workdesk version. But we'd like to know whether someone else is involved
> with workdesk use, usecases, implementations, etc.
>
> Should we have to develop it, refactor it?
>
> I'll be looking forward to your response,