Agreed, but it no doubt depends on the specific route and capacity on that route. Using Quic DNS it was also coming from Sydney but a different route, and much faster.

Hi all,

I backup data to my parents house in Hamilton on spark (900mbit plan) as a form of offsite for 3-2-1 backup strategy.

When I first joined quic I would be able to upload to them at 100mbit (i.e 1/5th of my theoretical upload speed), this was about 6 months ago. Since then the speeds gone down alot with to 20-30mbit with no changes to hardware either end. The performance on My republic before 2d took over was also around 150-200mbit.

I don't quite understand the jargon about route reflectors but do have a basic understanding of peering. My question is based on changes yet to come will things likely improve on the new network?

I've been quite enamoured by the journey quic is taking and I'm happy to stay if the roadmap may help with my issues. Otherwise it may be time to part ways (no I'll feeling) and come back once things have built out and settled.

Thanks for responses in advance.

Wait it out for the next two announced outages, and see how that improves things

theviper: I backup data to my parents house in Hamilton on spark (900mbit plan) as a form of offsite for 3-2-1 backup strategy.

 

I don't quite understand the jargon about route reflectors but do have a basic understanding of peering. My question is based on changes yet to come will things likely improve on the new network?

 

I've been quite enamoured by the journey quic is taking and I'm happy to stay if the roadmap may help with my issues. Otherwise it may be time to part ways (no I'll feeling) and come back once things have built out and settled.

The basic problem currently is Spark peering is really expensive (more expensive than some international transit) and when Quic stood up their new network with new peering etc they couldn't justify peering to Spark (and wearing that extra cost) on it at the beginning while they've got a perfectly good link already there on their old network meaning currently on their new network only traffic over to Spark is traveling over to Australia (where Spark openly peers).

I think the plan is they're about to move the spark peering across to their new network once they've migrated more of their other (Vetta) customers across to it. Quic was basically "early access" onto the new network just to get reliability back and take load off their old network to give them more breathing room. They're working on this as quickly as possible as running 2 networks is also really expensive.

It'll improve dramatically over the coming weeks. My employer is on a Spark connection so I fully understand! Working from home for me currently can be a little bit laggy at times! You can see on my Auckland Speedtest graph that for me it is 60ms away when normally it'll be around 8ms:

You could however change your parents from Spark to basically any other provider and it'll sort this issue ;)

Thank you for your thorough and detail reply. Hopefully by end of Feb it'll all be water under the peering bridge.

bet quic wished smokeping had never been born lol

below is someone else's smokeping but looks similar to most of mine, the latency seems to have got worse sometime between 22nd and 23rd?

I can see similar on a single one of my smokepings, also to a public DNS provider (albeit a different one) but the rest are fine. 

@pc I know that’s my Smokeping and you can’t take it too seriously at this time as I have a bunch of backups that run. 

michaelmurfy:

 

@pc I know that’s my Smokeping and you can’t take it too seriously at this time as I have a bunch of backups that run. 

 

your backups take 3 days? lol

was just an example but most of my graphs (below graph is to sydney linode vm) are showing the same increase from the same time, but maybe nothing/temporary etc etc.

@pc Ah, yes, that increase. I believe one of their primary paths died and so some routes are a bit scenic currently. Don't have any more information apart from that however. Was more saying my Smokeping graphs can't be all too trusted at this time as normally my server saturates my upstream for a bit that can affect Smokeping. I should really just bump up to Hyperfibre for more upstream bandwidth.

I'm sure @quic will be all over it tomorrow or at-least be able to comment a bit more than me. There are a couple we've noticed (they appear to be going via Cogent at the moment).

michaelmurfy:

 

@pc Ah, yes, that increase. I believe one of their primary paths died and so some routes are a bit scenic currently. Don't have any more information apart from that however. Was more saying my Smokeping graphs can't be all too trusted at this time as normally my server saturates my upstream for a bit that can affect Smokeping. I should really just bump up to Hyperfibre for more upstream bandwidth.

 

I'm sure @quic will be all over it tomorrow or at-least be able to comment a bit more than me. There are a couple we've noticed (they appear to be going via Cogent at the moment).

 

Any update on this @quic?

theviper:

 

Any update on this @quic?

Appears to be multiple fibre cuts due to conflict in the EU: https://ecogent.cogentco.com/network-status 

Welcome to the Cogent Communications status page. Customers Transiting our network from Europe to our Asian markets may be experiencing Latency and Packet Loss. A vessel is currently being prepared to carry out the operations from the submarine faults procedure. NO ETA.

It's affecting many providers (as well as upstream providers) currently (not just Quic) and tonight, hopefully during their maintenance, Quic will be bringing in another upstream provider and doing some network changes to help route around this.

Remember, upstream routing problems happen to all providers. It's just the nature of the internet.

theviper:

 

michaelmurfy:

 

@pc Ah, yes, that increase. I believe one of their primary paths died and so some routes are a bit scenic currently. Don't have any more information apart from that however. Was more saying my Smokeping graphs can't be all too trusted at this time as normally my server saturates my upstream for a bit that can affect Smokeping. I should really just bump up to Hyperfibre for more upstream bandwidth.

 

I'm sure @quic will be all over it tomorrow or at-least be able to comment a bit more than me. There are a couple we've noticed (they appear to be going via Cogent at the moment).

 

 

 

 

Any update on this @quic?

 

Planned maintenance has been completed last night to assist with routing around physical fibre cable issues with an upstream provider in various locations. Give it another try now.

quic:

 

Planned maintenance has been completed last night to assist with routing around physical fibre cable issues with an upstream provider in various locations. Give it another try now.

 

hasn't helped my graphs but not discounting it's something local if you think the quic network is all good. below is a 1 minute resolution smokeping type graph and a traceroute to that host. cheers

[pc@snakedog:~]$ mtr -wr4 linode
Start: 2024-02-28T09:56:14+1300
HOST: snakedog.kaseus.net               Loss%   Snt   Last   Avg  Best  Wrst StDev
  1.|-- _gateway                           0.0%    10    0.5   0.5   0.3   0.6   0.1
  2.|-- bng1-akl1.vetta.net                0.0%    10   14.3  12.7  11.0  14.4   1.3
  3.|-- xe1-2100-147.pe2-akl2.vetta.net    0.0%    10   14.2  12.6  11.1  14.2   1.1
  4.|-- xe1-2100-501.core1-akl2.vetta.net  0.0%    10   14.3  13.4  11.4  15.9   1.4
  5.|-- xe2-2700-110.core1-akl1.vetta.net  0.0%    10   37.4  36.2  34.6  37.8   1.1
  6.|-- 63949.syd.equinix.com              0.0%    10   37.4  36.2  34.4  37.9   1.0
  7.|-- 10.216.32.2                        0.0%    10   37.5  42.6  35.8  78.3  13.2
  8.|-- 10.216.35.6                        0.0%    10   37.2  35.9  34.4  37.5   1.0
  9.|-- 10.216.0.84                        0.0%    10   36.7  36.3  34.5  37.9   1.2
 10.|-- linode.kaseus.net                  0.0%    10   36.7  36.3  34.6  38.0   1.2