Hence the major problem with ec2- Cost. Yes, you can instantly scale up to a less powerful server than you would have got in the first place from a dedicated provider.

Also, You do know you can do things like upgrade the processor and ram on a dedicated, usually within a hour also. You just send in a ticket.

What server provider makes you do a contract?

That is simply false. No idea what kind of beef you have with amazon but please get your facts straight.

You seem to consequently ignore that with amazon you not only pay for a piece of hardware but also for the flexibility and infrastructure that comes with it.

Yes, your cheapo host might give you a gig more RAM and a few more megahertz for your dollar - but it doesn't have equivalents to EBS, S3, CloudFront, instant provisioning and all the other goodies that make EC2 worthwhile.

It will probably also not give you means to recover within minutes when your shiney, dedicated pizza box decides to blow a fuse.

Most servers sold from either provider come with 1-2TB of bandwidth. How much does 2TB of bandwidth cost on amazon again? Right.

I already said /multiple times/ that the other amazon services are well worth it. I only said EC2 /in my mind/ is not worth it, unless its for your overflow. If you have a static need for a server- As in, a webserver that will always be on, Then you are better off getting a dedicated. The bandwidth will cost less. The server itself will cost less. You will get more processing power.. You will get more RAM.

You will not get instant provisioning. But then again, I dont randomly change the specs on my webservers now either way.

340 dollars. If 340 dollars matter in your equation then I think we're talking about different things here.

I only said EC2 /in my mind/ is not worth it

Actually you come across like talking ultimo, just like in your next sentence...

If you have a static need for a server- As in, a webserver that will always be on, Then you are better off getting a dedicated.

See? That sentence simply doesn't work in that general form. If you have a need for a single webserver then yes, EC2 is probably not cost effective (although it can be even there, depending on the use-case). Once you're talking about a deployment with multiple servers the picture changes. Not always in the favor of Amazon but often enough to refute such a general statement.

Looking at it again it seems like all your arguments come from the angle of renting a single server. - In that light you make some sense. But please also see that amazon starts to make a lot of sense in the mid-range and up, in many situations.

I did a rundown a few posts up where I price out the need for a fairly common php/mysql website that would cost 600$ a month on ec2 to run, yet it only costs 200$ on dedicated.

Also, Im not talking about single server setups only. Show me a scenario where its cheaper to run your entire business on EC2 than a mix of Dedicated + EC2.

Seriously, I want to know of one situation where it would cost LESS money to run my entire 24/7 business on amazon then on dedicated or a mix of dedicated and EC2.

Thanks for pointing that out, my bad. Edited it as to not collect karma for false figures. I stand by my point, be it $140 more or less.

Also, Im not talking about single server setups only. Show me a scenario where its cheaper to run your entire business on EC2 than a mix of Dedicated + EC2.

Please see my first comment in this discussion where I said exactly that. It's exactly this mix which makes the most sense in most cases.

Also, Im not talking about single server setups only. Show me a scenario where its cheaper to run your entire business on EC2 than a mix of Dedicated + EC2.

I never claimed that - but yes, even those scenarios exist.

If your business is number crunching then EC2 might be a good deal. Likewise if your business is strongly saisonal (e.g. ticket sales for sports events or such).

Moreover there are many scenarios where EC2 enables things you could not realistically afford otherwise; do you think dropbox could have realized their storage and scalability needs on conventional hardware?

While traffic might be "ridiculously" expensive on AWS, storage happens to be riciculously cheap. You don't get $.10/GB "infinitely" scalable random access storage elsewhere.

And finally, closer to your usual PHP/MySQL shop, in many companies the savings due to flexibility and availability simply outweigh the markup on the hosting bill.

You keep throwing figures in the 3 digit range. Sorry man, but your admins burn that per hour while they reinvent all the little infrastructure things that amazon already has.

Trust me, not all of amazon's customers running those 1mio-or-something instances are idiots.

Im talking 3 figures because thats what I assume most startups need for their twitter/facebook/whatever app. It only becomes even more ridiculously expensive to stay with EC2 once you get bigger. You cant just say "well, once you are big its better!" Its not. Once you get bigger, the downfalls of relying on ec2 (for anything besides overflow) are even worse. The more servers you have and the more bandwidth you do, the more cash you throw out the window.

In fact as you get bigger you would be irresponsible to your shareholders to rely on amazon and not hire your own engineer. You know why? Its cheaper to have your own engineer and manage xx servers on colo then it is to have amazon. Not to mention you will still need a server guy once you get to that size anyways. Theres a reason you dont find any sites in the top 100 running their businesses entirely off of amazon. Its not financially prudent. Its a easy way out and not the /right/ way out.

Also, we are not talking about s3, simple DB or CDS, so stop bringing that up. We are talking about EC2. You dont get to say that EC2 is awesome because S3 exists. They are separate products.

I said multiple times that 1-3 server shops are irrelevant for this discussion. At that scale you can use anything and it simply doesn't matter either way.

Since you seem to have mild reading comprehension issues I will spell that out for you: Neither will a profitable business break a sweat over paying 500 more or less on hosting. Nor do many of the benefits of EC2 come into play at such a small scale.

So either throw out some stats and numbers or stop bullshitting people. There are enough people blindly drinking the cloud hosting koolaid without the fake facts.

I don't like your tone. I explained four possible use-cases in my last comment, I don't think I used "fake facts" anywhere. If you are immune to basic math then that's your problem, no need to get personal.

Its cheaper to have your own engineer and manage xx servers on colo then it is to have amazon.

I conclude that you simply have no idea what you're talking about, much less about the figures involved. Furthermore an EC2 instance apparently killed your pet dog at some point in the past, which now prevents you from any rational thought in that direction.

Upgrading hardware involves more downtime, and it's not anywhere near as easy to scale back down if I need to.

It only starts sucking when you actually have to serve files and do it quickly.