Hacker News Reader

Comment by sdze | Hacker News Reader

Why would anybody in the right mind centralize his/her infrastructure?

I doubt that people actually need something like Cloudflare.

I couldn't run search.marginalia.nu without it. I've seen up to 50,000 bot queries per hour (and peak out at about 500 human queries per hour). I don't have the hardware to cater to the bots. I also don't have the money to buy the hardware to eat the cost. The options are hide behind cloudflare or shut down the service.

It's not about traffic costs, but processing power.

Can you please explain what exactly bot were doing? What was their goal? Yes, I've seen bot scraping sites, which is expected. But what queries bots were doing towards niche search engine?

Search queries look like spam, like the sort of spam keywords you will find in comment spam. "Free cialis 50mg online pharmacy near me"-type stuff

Best guess is they're gambling I'm backed by Google's API and trying to poison their suggestion data.

Sorry I don't follow. Could please elaborate. You mean bots do query 'cialis' to get an ad-sense ad, while they are the same guys benefiting from ads shown? Or what? I genuinely want to understand the problem and most importantly the motivation.

I don't understand the motivation either, but I think what they are attempting is to make e.g. typing cialis into Google suggest specific queries like the one i showed, which may be so overspecified they provide the spammers' links.

That's my theory anyway.

Because they save us ~20 thousand a month in bandwidth cost.

We got 10k visits in a single day. Cost of data transfer: zero

Really zero, like non of the visitors was hitting the original servers? That would be impressive then. And you should consider to make money with delegating the traffic, not give away the traffic for free.

I mean, I did not went into the rabbit hole of checking thoroughly, but in cloudflare it says we served 8gb and aws says we served just a few megabytes.

You configure to ignore everything, even the url querystring, and worst case scenario, they serve your site from an internet archive snapshot. You can literally power off your server and the page stays online

Companies centralize their infra - be it on AWS or some VPS provider.

Was at a startup that paid 5k/mo for Cloudfront and moved to Cloudflare and paid just 200/mo. DNS performance improved as we switched over to Cloudflare as well. Saw a decrease in bot traffic. No complaints about usability or being blocked.

So yes, Cloudflare was useful and helped saved $ for us

You can use shit slow language with fat framework and just put it behind CF and run half decent, that's why people use it

But you can also use a fast language with no framework, but host it on a 5€/mo VM and put it behind CF and it will run half decent.

Made an account to say “their”. We are talking about cloudflare and cloudflare users. Their gender is not relevant in this conversation. At the end of the day we are people.

im non binary

Reply on news.ycombinator.com