It happens, but if I know in advance I can prepare for it. In my environment I handle this through hardware, software, or just plain communication(staged releases for various social networks/geographic regions). But I'm monitoring traffic before expected peak times, working with staff to monitor social media traffic(where they're advertising). I didn't say this wasn't difficult and/or not expensive, what I was trying to say is specifically with website traffic it shouldn't be unexpected or incapable of being planned for. If a company is launching a product and hyping it through social media, you should be actively preparing for high traffic. This is what you're trying to set up, so be prepared. It's not a good business practice to set customers up for failure. I could have looked at social media traffic, pre-order traffic, etc. etc. and increased by "x" percent to get a ballpark figure of estimated traffic. Now whether you want to or can do anything about it is a different story.jimmyjohnjohn wrote: ↑Fri Feb 07, 2020 9:55 amI appreciate the responses and will respectfully address them. I think everyone who responded are not accounting for some challenges which must be overcome in order to have a relatively successful release online release.
It doesn't matter if sprints and exclusives are new or not. A resilient system capable of handling load spikes 6 to 9 orders of magnitude from the mean will still be difficult to implement successfully and the fact that exclusives have existed for a period of time does not suddenly lessen that engineering challenge. The demand, apparently, is that this vendor should engineer a resilient system capable of handling spikes in usage that are 6 to 9 orders or magnitude higher than average that occur 0.0002 of the time. It is certainly possible to design and build such a system, but the cost to do so will quickly go parabolic, which is what you seem to demand the vendor implement in rather cavalier fashion. Again, I create distributed, scalable systems for Fortune 100 companies because they are the only ones who can afford it. Just to get your foot in the door is high 6 figures and these projects can easily reach 7, 8, 9, even 10 figures.
eRoc, what would happen to your environment if, for 0.0002 of the year, your load spiked by a million times where you were flooded with requests competing for read/write access to a single database record? Is your system designed for that? I can go on describing the process, but unless the system is designed to scale and ensure system state is protected, things will get messy quickly and the entire system may become unusable. Designing a system to handle the load that will be experienced 0.0002 of the time is a very costly proposition for a company and is completely outside the realm of expertise for a knife shop. It can be done, as you know, but it's expensive and it's always going to be difficult.
Cloudflare is a content distribution network. There is no way that Cloudflare alone would solve the problems that the vendor experienced. I don't want to be too critical but that shows a lack of understanding about the problem.
I don't really care how he runs his business or website. I got in on the 2nd pre-order no website issues at all. Maybe he could have continued with random drops, have a lottery system, email sign up, raise the price of the dang knife. Sometime you have to get creative.
Managing supply and demand has been a job since....thousands of years before websites. People are always going to complain, but if you set up customers for failure, the chances of them returning for business drops significantly. I didn't have a problem so I'll be back, but others obviously feel differently.