I'm moving mirrors of my busy-ish website from my hand-crafted dedicated colo solution into the cloud to try to get geographically closer to my global user-base, reduce latency and improve perceived performance, save money, and hopefully make administration easier. In part 2 of this series, I managed to get a minimal fairly " …
Hold on - you're worried about latency and interactivity speed enough to "justify" internationally-hosted servers but then you use virtual machines on a cloud infrastructure? Eh?
An international ping should never be more than about 100-150ms for anyone who's actually intending to make use of the Internet and it's so tiny an interval that download speed etc. are much more of an issue than how long a command takes until first response - you're not Google and don't need sub-microsecond response times. If you do, then a cloud-hosted VPS is about the worst possible thing to use because there is no guarantee of what the latency will be at all (for a start, you have no idea where the server is actually located in the target country, or what bandwidth is available, or what portion of CPU it gets, or when demand is going to go through the roof on nearby servers etc.)
Surely you would have been better off just getting a handful of smaller dedicated servers in this country? It would have been cheaper, I think, they would have been better specced, you would have more control over them, there would be no implications of foreign / international legal systems, and you wouldn't have any more latency issues than someone abroad pinging a site that doesn't use international servers (e.g. the BBC which is hosted almost entirely in the UK, I believe). And if someone like the BBC doesn't need local servers, it's doubtful that you do.
Also, your bandwidth could easily be sitting on a bog-standard 100MBps lines that they have running through most datacentres to supply racks and thus make your bandwidth / spike / response time / billing worries moot. You also won't be getting a bill next month of ten times the price because you hit an infinite loop. The "overall" traffic figures you wanted came out to about 3-4MBps constant - obviously there will be peak periods and spikes but that's not huge by most dedicated-server standards.
It seems to me to be a cloud advert (and although you mention problems with the very first step with some providers, you neatly gloss over them) rather than an article. If you're seriously considering running with 500Mhz and 512Mb, then you are vastly over-estimating the CPU power needed too - a bog-standard, bottom-of-the-line dual-core server could probably run 8 or 9 of those without too much trouble - and that's if you just moved those same virtual server images onto a single dedicated server. If you're paying £250 a month at the moment for a single colo, you should really being upgrading that hardware and/or instead buying 4-5 cheaper, smaller, dedicated servers from a variety of hosts (and then hardware failures don't need to factor into your costs at all).
With proper load-balancing I see no reason why you would need to own a single byte outside of the country you're in. It just seems odd, like every other "justification" for cloud servers that I can imagine.
Your basic assumption...
Is wrong. And that is because you don't know what this guy's needs are.
Neither do I, for that matter. And that's the main shortcoming of the article. Otherwise, well done.
Have you looked a these guys
- Nokia: Read our Maps, Samsung – we're HERE for the Gear
- Kaspersky backpedals on 'done nothing wrong, nothing to fear' blather
- Episode 9 BOFH: The current value of our IT ASSets? Minus eleventy-seven...
- Too slow with that iPhone refresh, Apple: Android is GOBBLING up US mobile market
- Analysis Uber, Lyft and cutting corners: The true face of the Sharing Economy