But the bots on my site -- at least the obvious ones that lead me to say they are a large source of traffic -- are all well-behaved, with good clear user-agents, and they respect robots.txt, so I could keep them out if I wanted.

I haven't wanted because, why? I have modified the robots.txt to keep the bots out of some mindless loops trying every combination of search criteria to access a combinatorial expansion of every possible search results page. That was doing neither of us any good, was exceeding the capacity of our papertrail plan (which is what brought it to our attention) -- and every actual data page is available in a sitemap that is available to them if they want it, they don't need to tree-search every possible search results page!

In some cases I've done extra work to change URL patterns so I could keep them out of such useless things with a robots.txt more easily, without banning them altogether. Because... why not? The more exposure the better, all our info is public. We like our pretty good organic Google SEO, and while I don't think anyone else is seriously competing with google, I don't want to privilege google and block them out either.

Bots would routinely try to scrape pricing for every combination of {property, arrival_date, departure_date, num_guests} in the next several years. The load to serve this would have been vastly higher than real customers, but our frontend was mostly pretty good at filtering them out.

We also served some legitimate partners that wanted basically the same thing via an API... and the load was in fact enormous. But at least then it was a real partner with some kind of business case that would ultimately benefit us, and we could make some attempt to be smart about what they asked for.

As described elsewhere, rate limiting doesn't work. The bots come from hundreds to thousands of separate IPs simultaneously, cooperating in a distributed fashion. Any one of them is within reasonable behavioral ranges.

Also, caching, even through a CDN doesn't help. As a B2B site, all our pricing is custom as negotiated with each customer. (What's ironic is that this means that the pricing data that the bots are scraping isn't even representative - it only shows what we offer walkup, non-contract customers.) And because the pricing is dynamic, it also means that the scraping to get these prices is one of the more computationally expensive activities they could do.

To be fair, there is some low-hanging fruit in blocking many of them. Like, it's easy to detect those that are flooding from a single address, or sending SQL injection attacks, or just plain coming from Russia. I assume those are just the script kiddies and stuff. The problem is that it still leaves a whole lot of bad actors once these are skimmed off the top.

[1] https://en.wikipedia.org/wiki/List_of_largest_Internet_compa...

Being on that list puts the company's revenue at over $1 billion USD. At a certain point it becomes cheaper and easier to fix the system to handle the load.

So your company is deliberately trying to frustrate the market, and doesn't like the result of third parties attempting to help market efficiency? It seems like this is the exact kind of scraping that we generally want more of! I'm sorry about your personal technical predicament, but it doesn't sound like your perspective is really coming from the moral high ground here.

No. First, we as a middleman resseller MUST provide custom prices, at least to a certain degree. Consider that it's typical for manufacturers to offer different prices to, e.g., schools. This is reflected by offering to us (the middleman) a lower cost, which we pass on to applicable customers. Further, the costs and prices vary similarly from one country to another. Less obviously, many manufacturers (e.g., Microsoft, Adobe, HP) offer licensing programs that entitle those enrolled to purchase their products at a lower cost. So if nothing else, the business terms of the manufacturers whose products we sell necessitates a certain degree of custom pricing.

Second, it seems strange to characterize as "frustrating the market" what we're doing when we cooperate with customers who want to structure their expenses in different ways - say, getting a better deal on expensive products that can be classified as "capital expenses" while allowing us to recover some of that revenue by charging them somewhat more for the products that they'd classify as operational expenses.

> it seems strange to characterize as "frustrating the market" what we're doing when we cooperate with customers who want to structure their expenses in different ways

I'm characterizing the overall dynamic of keeping market price discovery from working as effectively. How you may be helping customers in other ways is irrelevant.

They’re trying to pay less tax by convincing your company to put a different price on products they buy based on their tax strategy.

It sounds illegal.

What do you think about their other assertion that the search page is getting a gigantic number of hits that a/ cannot be cached and b/ cannot be rate limited because they're using a botnet?

The scale of the botnet sounds like an awfully determined and entrenched adversary, likely arising because this company has been frustrating the market for quite some time. A good faith API wouldn't make the bots change behavior tomorrow, but they certainly would if there were breaking page format changes containing a comment linking to the API.

The thing I still don’t understand is why (edit server not cdn) caching doesn’t work - you have to identify customers somehow, and provide everyone else a cached response at the server level. For that matter, rate limit non-customers also.

Search results obviously can't be cached, as it's completely ad hoc.

Product details can't be cached either, or more precisely, there are parts of each product page that can't be cached because

* different customers have different products in the catalog

* different products have different prices for a given product

* different products have customer-specific aliases

* there's a huge number of products (low millions) and many thousands of distinct catalogs (many customers have effectively identical catalogs, and we've already got logic that collapses those in the backend)

* prices are also based on costs from upstream suppliers, which are themselves changing dynamically.

Putting all this together, the number of times a given [product,customer] tuple will be requested in a reasonable cache TTL isn't very much greater than 1. The exception being for walk-up pricing for non-contract users, and we've been talking about how we might optimize that particular cases.

The low millions of products also makes some sense I suppose but it's hard to imagine why this doesn't simply take a login for the customer to see the products if they're unique to each customer.

On the other hand, I suspect the price this company is paying to mitigate scrapers is akin to a drop of water in the ocean, no? As a percent of the development budget it might seem high and therefore seem big to the developer, but I suspect the CEO of the company doesn't even know that scrapers are scraping the site. Maybe I'm wrong.

Thanks again for the multiple explanations in any case, it opened my eyes to a way scrapers could be problematic that I hadn't thought about.

I haven't had to defend against extensive bot scraping operations -- only against simpler ones -- but I've utilized such a practice in my admittedly much more limited experience, and was actually successful. Not that the bots gave up but their authors realized they can't accelerate the process of scraping data so they dialed down their instances, likely to save money from their own hosting bills. Win-win.

Apologies, I don't mention to lecture you, just sharing a small piece of experience. Granted that's very specific to the backend tech but what the heck, maybe you'll find the tidbit valuable.

It's been forever since I worked at Yahoo Travel, but bot traffic was significant then, I'd guess roughly 5-10% of the traffic was declared bots, but Yandex and Baidu weren't agressive crawlers yet, so I wouldn't be terribly surprised if a site with a large catalog that wasn't top 3 with humans would have a majority of traffic as bots. For the most part, we didn't have availability issues as a result of bot traffic, but every once in a while, a bot would really ramp up traffic and cause issues, and we would have to carefully design our list interfaces to avoid bots crawling through a lot of different views of the same list (while also trying to make sure they saw everything in the list). Humans may very well want to have all the narrowing options, but it's not really helpful to expose hotels near Las Vegas starting with the letter M that don't have pools to Google.

Pretty much Twitter and the majority of such websites.