EC2 performance (and other virt performance) is absolutely abysmal compared to dedicated hardware, particularly with respect to I/O. It's trivial to get a ton of IOPs into your own hardware and expensive/hard to get similar numbers out of EC2.
Secondly, if your capacity varies hugely, it's likely you can engineer a 'scale into the cloud' solution where your baseline capacity is your own hardware, and extra queues get spun up in the cloud to deal with unusual peaks.
It really sounds like you guys are way above the capacity levels where cloud solutions make sense.
Even if you build something great, and give it away, it'll still take time, effort and money to let people know that it exists and that it exists, that they should care, and that they should be impressed.
If you build it, nobody will care. They won't come.
If you build it, flog it for ages, and shove it down your market's throat, they might come.
I also tend to take it on face value, because my experience is that women are treated extremely poorly at these events, particularly if they're even a little bit attractive.
Large numbers of men, many of whom have social issues, widely available alcohol, and small numbers of women... it's a recipe for something really terrible.
but with lack of proof (and that certainly seems to be the case by reading the blog post) it is a very slippery slope to name the person on your blog.
Let me get this clear: it's your opinion that unless somebody videotapes every moment of their life, that they should not publicly state negative things that happen to them?
edit: for the downvoters, which part of that question is inappropriate?
For what it's worth, nearly every major firm I've worked with rolls their own metrics.
Mixpanel and all these other metrics startups are cute, but at this point in time the risk with integrating MixPanel is almost certainly viewed by most managers as greatly exceeding the risk associated with rolling your own metrics.
Secondly, if your capacity varies hugely, it's likely you can engineer a 'scale into the cloud' solution where your baseline capacity is your own hardware, and extra queues get spun up in the cloud to deal with unusual peaks.
It really sounds like you guys are way above the capacity levels where cloud solutions make sense.