I’m switching to a new ISP that has PPTP tunneling for better routes for gaming. I can apply this at router level except that the tunnel has a 10Mbps limit per connection so as a result I’ll need to split tunnel and define destination addresses / domains that should take the PPTP tunneling route.
TL:DR
On a domain level. What are the domains WoW uses and does Blizzard rotate IPs for load balancing? If yes then this would not be a viable option to route based on domain.
Or if the above is not known, what are the remote IPs for WoW? I saw some information in one of the support articles: https://us.battle.net/support/en/article/7870
Though when checking the router log I dont see any of these IPs listed. But if the IPs rotate then once again this would not be a viable option.
I don’t believe that Blizzard discloses any of its network infrastructure information, owing to the fact that while it would make it easier for users to configure their own routing, it also makes it easier for attackers to cripple their network infrastructure.
But theres no harm in asking, they can always say “no”, I guess.
If your ISP has a better route for the tunnel, why does it not use the better route to Blizzard to begin with? Also the tunnel overheads will reduce MTU and require additional processing. To say nothing about the extra configuration that is a large source of failure.
Tunnels are not actual tunnels. You do not shortcut to the destination because it has to take the exact same path. Unless your ISP is doing something very suboptimal, in which case why are you using them?
Blizzard are also not going to give you their infrastructure details.
Not sure where you got the 2-3ms delta. I’ve not yet moved to this ISP. The ISP itself is the only one that offers 1Gbps and this is an option that they offer. I’m simply exploring potential options from my side and I intend to test them. This is in no way a requirement for me.
Regarding the potential security vulnerability, that’s why I’m planning to essentially “whitelist” which traffic takes this route i.e only game traffic configured directly on my router.
I’ve not mentioned that this is the only ISP that offers 1Gbps for me so that’s my motivation for moving to them. Regarding tunneling overheads, yes they exist but in what capacity in terms of impact we don’t know exactly especially when said tunnel is run directly from the router in my case which has dedicated hardware for optimizing this. Hence the need to test to see exactly what the numbers are. Right now this is all experimental and I couldn’t care for 10-20ms difference in latency even. This is an interesting scenario I’d like to test is all.
You are not going to escape MTU overheats, and I repeat the only way this would improve routing is with suboptimal routing already there.
A network tunnel is not a shortcut through the ground. It is not a case of a tunnel is 10km but a road around is 30km. The tunnel will be 30km +2km for the tunnel ends, running parallel with the road.
Unless your ISP is making you go 30km around instead of using the 10km route which is already there for no reason, when every other ISP would use the 10km route because it gets you off their network more quickly (saving them money* and making you happier). In which case they are a bad ISP.
I work in this field and me and my colleagues were honestly WTFing at this, and wondering if your new ISP was trying to scam you or something. Because it’s utter nonsense.
Technically you could have stuff like transit you have to pay for and a company could force a longer way around to ‘free’ peering, I mention this as a pedantic aside.
The OP has probably seen all the hops his packets take, and thought that he could do better, routing them directly from endpoint to endpoint, even though there actually isn’t a direct connection from endpoint to endpoint (as in no physical connection from him to Blizzard).
They don’t seem to realise that packets can only be directly routed to places that actually have a physical connection.
Seems like you made up an argument in your head and then proceeded to defend said obvious head argument. This is not me thinking I can do better. This is a recommendation from the ISP and is an article listed by one of their engineers. The point to point tunnels itself belong to the ISP as well and apparently get directed to routes over Seacom. Once again I don’t claim this is a practical solution nor the best solution. I simply am pursuing a what-if scenario to test based on their recommendation, which honestly cant hurt.
With regards to suboptimal routing. That is admitted by them in the very same post hence the need for the tunnel. Once again this is just my preparation for something I may not even experience at all. I do believe you are viewing routing within a relatively local sense in this scenario. I am roughly 13 000km away from EU servers so this is addressing local routing issues before handoff to international underwater sea cable (according to them). There’s currently 3 tunnels, one for each major city in the country all posts have reported lower latencies however the logic and practicality of this I honestly couldn’t care as much for, but just to view this as an experiment as a whole.
Yes they are getting a metric butt-ton of flak in the thread because like my current ISP they could simply classify this “gaming” traffic and redirect it over their supposed optimal route.
I worked at a major European ISP with a global presence for years, and have dealt with many complexities with Internet routing. I have never heard of anything like this, and would be delighted to read the article if only because my colleagues wouldn’t mind a good laugh.
If they allow you to point specific traffic over the Seacom submarine cable instead of dumping all their transit to Europe on there ASAP and make ‘normal’ traffic go the long way around then this is an awful ISP cheaping out on their transit, and pushing a lot of complexity to the end customer.
Completely agree with you. It’s absolutely deplorable service as a whole to deal with an issue with a half-baked solution (if you can call it that) and then to hand that over to the customers. Absolutely abysmal.
But really the concept of split tunneling in this manner is interesting and I’ve not been in a scenario where I had to do this on my home network and for that low-brain reason alone I’m very much intrigued to see what happens.
My current ISP does offer the best experience in terms of latency compared to all other local ISPs and you’ll also find their packages sold at the highest premium of all other ISPs. They just don’t have what I’m looking for hence my choice to compromise latency a bit for throughput with this new ISP…
I’m probably still going to test this just to see the numbers. I do see some of the remote IPs and domains used when logging in and running around. I’ll just try to scope them in and see if numbers go brrr.
[UPDATE 29-06-23]
Testing the above. There’s a 40ms improvement in latency and overall feel is significantly better. Though I am routing all internet traffic over the PPTP tunnel for this test which is quite risky but just for testing for now until I can get all traffic with “x” remote port ONLY to use the PPTP tunnel.