Evaluate best way to reduce the task data for process instances

[harmeet-status]

We should look into these 2 things:

1. Find a way to only store changes to the task data, incremental approach
2. Compressing all the task data server side
3. Anything else?

[harmeet-status]

Let's chat on Wednesday morning and discuss progress.

[burnettk]

jason rocked out the compression.

it was discussed that bpmn process architects tending process data and process instance archival are generally good practices that should be continued and discussed as we go forward. we also were thinking that possibilities for other set-it-and-forget-it code solutions probably still exist, such as:

1. store top level keys in separate documents
2. store diffs instead of full documents, and maintain current data snapshots for performance, potentially via non-task-data based environment

we should probably pick one of these approaches and prototype it in order to determine if it will be a complete solution or whether the complexity is not worth it (and we should stick with compression along with non-code solutions for the time being).

@harmeet-status will file a new ticket for implementation.

[harmeet-status]

@burnettk we dont need to file a new ticket, we agreed to do a POC. Let's do it as part of this ticket?

[harmeet-status]

What's the latest on this ticket? @burnettk

[burnettk]

Jon’s hammering on the first part of it. We think there’s a good chance we can reuse some code related to a non-task-data-based environment
to help here. More soon!