Hi folks!
Camunda 8 documentation mentions that payload of Process instance is up to 4MB, and it cautions to store much data in the process context: Variables | Camunda 8 Docs
Could you please provide some similar reference to how things work in Camunda 7. From a recent question, Process Variable size limitation
It is apparent that size of JSON perseus variable values can be 4GB if the database is MySql. (Size limit of longblob type is 4 GB in MySql). So what would be the performance impact if we have multiple variables of large size?
Really appreciate help and clarity. Thank you!
Hi @akangupta1, I can not find it right now, but there was a blog post on camunda blog about the impact of variables on the performance with detailed tests and some fancy graphs of performance to variable relations. They key takeaway was: with each variable, the execution time increases exponentially, therefore the good practice is to avoid storing complex data in variables. Proper solution to this kind of use case is to create a dedicated storage space (dedicated entity table in DB or cloud storage or any other) and in variable keep only reference to stored object, this way you limit the amount of variables and improve performance as large variables do not have to be loaded every time (especially if you are trying to access them from workers)