[INFO]: network graphs 0.1.9 vs 0.1.10

[zoff99]

No description provided.

[zoff99]

v0.1.9

[zoff99]

v0.1.10

[nurupo]

Nice. This is not the final result though, based on what @zugz has said in #542, we will start to see the difference once majority of the network switches to 0.1.10, which didn't happen yet.

Edit: mentioned the wrong person

[zugz]

Interesting. Thanks. Was this using UDP or a TCP relay? How many offline friends did you have? Is 0m0s when you started up tox in each case?

[zoff99]

tcp relay. 4 friends (online)

[zugz]

* Saturday, 2017-08-26 at 02:49 -0700 - Zoff <[email protected]>:

tcp relay. 4 friends (online)

OK. If you're using TCP, so not forming part of the onion, it shouldn't matter what version other people on the network are using. But are these graphs from soon after you start tox up? The bandwidth use should go down significantly after a few minutes of operation, once announce nodes become considered "stable".

[zoff99]

about 10 minutes in

[zoff99]

anybody with good knowledge about tcp relay code get in touch with me please
@zugz i want to tweak the tcp relay code to make it more mobile friendly, but need some help

[zugz]

* Saturday, 2017-08-26 at 10:54 +0000 - Zoff <[email protected]>:

about 10 minutes in

OK, then I believe almost all the traffic you're still seeing in 0.1.10 should be DHT traffic. I expect that could be reduced quite a bit too.

[zugz]

Actually no, I've idea what the traffic is...

[iphydf]

@zugz what is the idea?

[zugz]

* Tuesday, 2018-01-16 at 11:55 -0800 - iphydf <[email protected]>:

@zugz what is the idea?

That was a typo - I meant "I've no idea what the traffic is"! That's still the case. But I think we can close this issue.

[ghost]

Whilst this is a great reduction what I don't get is why don't we use a reactive system where a node coming up sends an update to all it's contacts and conversely sends a down update to all it's contacts and does no polling other than if it has a message in it's out queue for a node that is not responding.. This would make a seriously large impact to idle traffic and cpu loads. What do you all think? I mean if all the polling is just to let someone know if people are online - which in most case we do not give a hoot about until we want to communicate with them, is a huge waste of data allowance and battery energy. I know that currently there is no outbound queue but yeah we need it big time....pls.