Running single instance of Elasticsearch in production

a couple of question related to production configuration:
a) We have lots of stuff on our box including Elasticsearch and we don't want it to bring down other systems, so we decided to list direct memory usage for Lucene, as I guess, with this parameter ES_DIRECT_SIZE=1g. It 1g of memory quite enough for 100 msg per second and data for the last 7 days?

b) We run single instance of elasticsearch 2.0 and we don't want it to cluster with other instances in the same network. I looked here:

"Elasticsearch 2.0 takes a different approach. Multicast has been removed (although it is still provided as a plugin for now). Instead, and only when bound to localhost, Elasticsearch will use unicast to contact the first 5 ports in the transport.tcp.port range, which defaults to 9300-9400."

now I'm confused: I bind elasticsearch instance in production to localhost with " localhost" parameter. does this say that my instance can join a cluster? how to prevent this??

I wouldn't recommend running ES with any less that 2GB of heap, even for dev.

If you have bound to localhost, then only other instances also running on that same host machine also running on localhost with the same clustername and the zen discovery list on both nodes for them to join together.
So I think you will be ok in that respect.