Twitter Streaming API Best Practices(Multiple or Single connection)

What is the best practice if I have 20K twitter user base and I want to track user's specific keywords via statuses/fi开发者_如何学编程lter?

Should I distribute the processing on multiple nodes, lets say open a streaming connection tracking keywords for 5K users each (on different IPs or same IP with different authenticating users)?

Or just apply for a bigger access level and use a single connection to get the whole thing.

Thanks, Alam Sher

Bigger access level. Using multiple accounts to circumvent access limits is

1. Frowned upon by Twitter
2. Complicates your processing