Anyway, I’ve been using that setup b/c it’s inside the AWS free-tier ; but with the problems I’ve had recently (see last post), I’m not sure if it’s the on-demand-iness, the micro’s virtual disk instead of an s3 distributed drive backing (my money is on the on-demand bit).

So I’ve been looking at the prices.  I looked first at the low-utilization micro instance, since I really don’t see enough traffic to qualify as “high utilization,” but apparently in AWS-speak, utilization is “how much would this system be online.”  Since the blog/sites would be on all the time, that’s “high utilization”

Current prices are on-demand at 2c/hr, so 1yr is 175$

Light Reserved is 23$ registratoin and 1.2c/hr, total 130$

Heavy utilization is 62$ and .5c/hr, 105$

To go up to a standard small (instead of micro) for the s3 backing:

on demand – 8.5c/hr, 745$

light – 98$ & 5c/hr, 535$

heavy – 276$ & 2c/hr, 450$

So … yeah, apparently it’s not quite as cheap as I was thinking for the year on AWS, but even if I split out across three instances for the various bits (my blogs, a couple sites we host, and a forum we run) it’d be ~300$/yr, opposed to currently we pay ~80$/mo on the business internet to run those things .  Still a net-savings of 400$ give or take.  And a massively higher throughput ; Not to mention that we can upgrade our home network from 1.5/1.5 business to probably 25/5 and STILL save 50$ a month  (there’s more on that bill, so I’m not sure which parts exactly would be removed, but I’m sure it’s more than the 300$ I’d be paying on the new structure)

So, there will likely be some flickering as I move to the reserved instance in the next month or so; I’m really hoping it helps the occasional choke-out I’m seeing.  If not, then I will DEFINITELY be putting in a support question

But it seems like maybe it’s having some issues with the 24/7 uptime concept, b/c the other day it was mentioned that my main blog had flaked and browser was declaring “unknown!”

Well, sure enough, couldn’t ssh into the server either, weird.  All the status checks said good, but the system was definitely not responding well.  So I bounced it from their admin interface, it came back up, and has been happy for a few days now.

So, I’m guessing that it doesn’t actually like being on all the time, maybe it swaps out when it’s not being used, and eventually gets GC’d even though it’s “active”

Not sure why the status notices said clear though …

Anyway, that’s a potential issue I’ll need to figure out better before moving anything bigger over.