125
ARCHITECTURE AND PERFORMANCE IN SDNs
© 2010 Taylor & Francis Group, LLC
quite well known: batching I/O to minimize its overhead, port-
ing the I/O handling harness to Boost asynchronous I/O (ASIO)
library (which simplifies multithreaded operation), and using a fast
multiprocessor-aware malloc implementation that scales well in a
multicore machine.
Despite these modifications, Nox-MT is far from perfect. It does
not address many of the performance deficiencies of Nox, including,
but not limited to, the heavy use of dynamic memory allocation and
redundant memory copies on a per-request basis, and the use of lock-
ing while robust wait-free alternatives exist. Addressing these issues
would significantly improve the performance of Nox. However, they ...