Real time flag

Stephen Oliver edited this page Apr 3, 2017 · 1 revision

The real-time flag, or bulk flag, was added in Fred 1311.

It gives us the choice of two types of requests:

Real-time requests

Optimized for latency, they are expected to be bursty, and each node accepts fewer of them in order to reduce worst-case latency, as well as limit resource usage.

However their transfers are prioritized over bulk requests (with an anti-starvation mechanism).

Bulk requests

Optimized for throughput, they are expected to be large and constant, and each node accepts a lot of them, making the worst-case transfer time very high.

Details

In the context of new load management, real-time requests should be lower queueing latency if they are bursty. However, because we accept fewer of them, this may or may not be true in practice.

If it proves to be absolutely necessary, we might allow routing to be compromised a little for realtime requests, whereas for bulk requests we can happily queue for as long as it takes within reason, even if that means 5 seconds per hop.

Clone this wiki locally
You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.