[wmf]

The sat-to-sat laser links are used to provide connectivity on the open ocean and in remote parts of Australia and Argentina that are beyond the range of any ground station. They're definitely working but AFAIK they are only used when necessary so if you're within range of a ground station your traffic will never use laser links.

[Fischgericht]

Oops, forgot one important thing: Sure, why do additional hops if you can see the base station. But what about shared state? Why do you definitely still get a completely new session when moving to the next sat? If the laser links are working, that state should be shared between neighboring sats.

[lxgr]

Inter-satellite links simply provide additional (time-variant) paths, which doesn't inherently relate to shared state.

You seem to be under the impression that inter-satellite links somehow imply a self-organizing mesh topology that preserves terminal-to-gateway associations at any cost (including that of extra in-space hops), but that does not necessarily follow from the existence of ISLs.

In other words, your observation of occasional routing instability causing higher-layer issues is perfectly compatible with working ISLs.

[Fischgericht]

Accepted.

[inemesitaffia]

Starlink switches beams every 15 seconds and satellites every 120 seconds.

You keep your sessions through both. Lasers or no lasers

[JumpCrisscross]

> Why do you definitely still get a completely new session when moving to the next sat? If the laser links are working

Imagine Amazon 10x'd its ingress/egress fees between regions.

[inemesitaffia]

You're not getting new sessions period.

[Fischgericht]

I will not disagree as I can not verify this claim. Have you tested it yourself or have a source which has some tech proof on that one?