Performance of TNonblockingServer, TThreadPoolServer for DB-bound server in Python

Question

I am making an internal API in Python (pardon my terms) that provides a layer over MySQL and Solr (databases) with only simple computing. A Python program that spawns from scratch waits 80ms for Solr, while taking negligible time by itself.

I am worried about the incomplete threading support of Python. So which of the modern Thrift servers allows high-performance request handling?

---

In Python, I could make a WSGI app under Apache workers that:

• pooling resources such as DB connection objects
• high performance with minimum processes
• graceful dropping of requests
• (relatively) graceful code reloading
• a keep-alive mechanism (restart the application if it crashes)

Answer 1

Apparently TProcessPoolServer is a good server and forks different processes, avoiding threading issues.