Hi,
We run jobs that trigger 1000s of commands for certain aggregates in one shot and as a consequence generate 1000s of events that eventually will update our projections (one view record per event). The whole processing takes quite a lot of time.
I wonder if there’s best practices that would allow handling such processing in batches?
Some ideas…
Are you using tracking event processors, or subscribing event processors?
If you are using tracking event processors could you scale out the number of segments/threads you have the processor split into?
Could you isolate this event so that it is in it’s own event processor? This may help with throughput as it is handling only this event.
Ben