I’m dealing with an app that is meant to consume thousands of events per sec. The particularity here is that those events are very, VERY similar to one another. They not only follow the same schema, their length is very similar, and if we put our backs into it, we could parse it as a positional string, since there are only a couple of strings that would alter the total length of the message…

Given this, and trying to avoid the positional parsing, are there any tricks I could do, as a deserializer, to make deserialization faster and/or memory economic?