Hacker News Reader

Comment by xorcist | Hacker News Reader

I fully understand why a company would like to launch this type of service. This is the free market after all, and it would make the company insanely valuable, should it succeed.

However I do have an issue with the marketing behind it. While not said outright, there is a clear message here that due to some unspecified magic your network performance will increase. That's clearly stretching the laws of physics, at the very least. There are also nebulous privacy statements which looks conspicuously like services that shield your identity, which does not seem to be the case here.

If the real intent here is to help underprivileged Internet users escape their great firewall, onboarding some regular users might be necessary to make the service more legitimate. However even a generous reading of this announcement does not seem to support this use case. The consumer VPN business is a questionable business at best, and this does not look different.

View on HN

> That's clearly stretching the laws of physics, at the very least

This is not clear to me. Few mobile users have pings to anywhere pushing up against the speed of light, and the bandwidth/loss/routing is not close to being limited by physics.

No, he's correct here, it's an extra indirection and so overall it can only hurt performance, not improve it. For example, ISPs often have caching servers from Google installed and Google is the biggest traffic generator. If you route traffic through somewhere else you are going to reach different Google's caching servers that are farther away and over more congested links.

> No, he's correct here, it's an extra indirection and so overall it can only hurt performance, not improve it.

Really? The cost of an extra hop is just one of tens (hundreds?) of factors that would determine if using WARP would be faster for a particular scenario.

That's not an extra hop, but essentially two different paths with many different hops instead of a more direct one. I'm sure there will be edge cases where this can be faster due to one of those factors, but for most destinations it will be slower.

In aggregate, is it necessarily an extra indirection? If, for example, your ISP bounces around a bunch getting to your destination and Cloudflare has a faster path, isn't it possible they have the faster path?

It was not meant in a literal sense, I hope that much was clear. The language was suggestive on purpose considering the article being commented on. A longer path takes longer time as speed of light varies comparably little, that much is physics.

And sending traffic by way of a third party is very likely to make for a longer path, for most people in most circumstances. Not by necessity but because few have that lousy routing agreements, unless we're talking about special circumstances such as 6to4 tunnels and the like.

What basis are you making this on, have you worked on a multi-homed BGP environment before? There is a lot of latency hiding in the route maps and interconnections most providers use, and running your own lines can reduce a ton of jitter, route instability, congestion, and other common issues where most tech people oversimplify the constituents of the Internet.

"due to some unspecified magic your network performance will increase"

In this post, they say that routing over cloudflare's network can be up to 30% faster because they maintain more efficient routing information than the public network:

https://blog.cloudflare.com/warp-technical-challenges/

They could speed up your international traffic (they don't tho).

Reply on news.ycombinator.com