We've discussed switching to Kafka. There are some pros/cons to doing that. With respect to my problem above, our messages _could _conceivably approach 1MB (or even surpass it), so we're really just delaying the inevitable. That said, we're a long, long way from hitting that limit, so it's definitely something we're looking at.
We just recently started gzipping our payloads, which buys us even more time.