COBOL has always been a high performance language within it’s typical user cases. So I’m not too surprised that it also performs well on other levels as well. But writing a framework to handle the encoding and decoding of bits and packets in that language sounds like a special kind of masochism.