Scribe, Flume and Chukwa - what are alternatives?

Question

Hello, I would like to learn about alternatives to those projects, especially designed to aggregate data from logs from multiple nodes (>500) and designed for low memory/cpu usage.
I'm familiar with scribe, flume and chukwa and I think that they use too much resources (JRE, many libraries).

All comments and recommendations about aggregating data from multiple nodes in low-usage manner are welcome.

Answer 1

"I'm familiar with scribe, flume and chukwa and I think that they use too much resources" --- can you please elaborate the above? AFAIK, Scribe is written in C++, and it's pretty efficient in resource usage.