I haven't worked with Talend Data Integration on real-time data processing; I have only seen the streaming component, and I think it can accommodate limited streaming, not at the level of larger big data platforms. Regarding ETL, Talend Data Integration is great, but concerning real-time data processing, people are not really sure about Talend Data Integration or might not know how it provides such types of flexibilities. I haven't worked on this aspect in Talend Data Integration, so this is definitely an area where it lacks. To elaborate on one drawback, if I am designing something in Talend Data Integration, I cannot backtrack in the flow. For example, if I read data from a source and transform it before applying complex aggregation in the third step, I cannot join this data with the second step. If I need to go back, I will have to use a memory component, or if I need to avoid memory consumption, I will have to read the data from step one again to join it with the third step.
