I run admin + staging stuff and my own email on the dedicated servers. The Hetzner box has some offsite backups, too. I want something that won't die at the same time as the website.
I use two colos for redundancy. One is in Sacramento, and it mitigates the earthquake risk of the one in Fremont.
The AWS costs consists of stuff I'm too lazy to go find and delete. Everyone has an unused EC2 instance or S3 bucket somewhere; that's what keeps Amazon in the black.
Interesting. Are colo prices generally that high in the US? May I ask what's included at 600$ for 2(-3?)U of rack-real estate there?
* What level of support do you have (can you call someone at the
DC at 4am and have him hot-swap a failed HD for you)?
* How many shared or own gigabit or 100mbit uplinks are included?
(if >1: can do bonding on your interfaces?)
* With how many other servers do you share the same circuit-breaker?
* I guess you have at least 1 UPS + 1 non-UPS power supplies?
* exceptionally good peering?
* network/power SLA?
* 24/7 access to your server?
I worked for a DC in europe some years ago and i'm genuinely intrigued how the price in your spreadsheet for colo-hosting is so much more than what i was used to (~100-150$/U).
> Note that 15 A is nowhere near enough power to fill it to even 1/3 capacity.
Yes that's pretty standard, afaik. The dc I worked at had individual circuits even for quarter- and half-cabbinets (2x 16A each), but i don't know how common that is in general. (instead of just putting 4 quarter-cabinets on the same circuits and have 1 customer blow his and 3 others' power)
Yeah, I still haven't made up my mind as to whether I should trust all my bookmarks with a guy who seems to have a knack for disappearing alone into strange corners of the world. Even if he is occasionally protected by a company of tanks and a Hind helicopter :)
I use two colos for redundancy. One is in Sacramento, and it mitigates the earthquake risk of the one in Fremont.
The AWS costs consists of stuff I'm too lazy to go find and delete. Everyone has an unused EC2 instance or S3 bucket somewhere; that's what keeps Amazon in the black.