- NGINX
ARCHITECTURE | STATUS |
---|---|
Non-ISO | OK |
Non-Portable | OK |
Synchronous | BAD (?) |
Single-Threaded | BAD (?) |
First-Come First-Serve | OK |
REST-FPM will request different data types simultaneously through libCURL's Multi Interface (lCMI). This means that a request for light data is handled in a separate thread than a request for another type of data.
Client connects to REST_FPM through FastCGI
1. REST_FPM parses request for N types of data
2. REST_FPM sends HTTP GET request for each piece of data simultaneously
3. Parses first request to complete
4. Parses second request to complete
X. [...]
N+2. Parses Nth request
N+3. Writes data to client
- Synchronous Parsing
- Mitigated only when data requests do not return instantaneously
- Others (?)
I do not expect a bottleneck to be an issue for our use-case. We will have very little unexpected traffic and the case of having to serve simultaneous clients is a non-issue. The architecture outlined should handle serving <= 50 clients with ease.