Is EC2 The Right Choice For Hosting Our Web Application? | Hacker News Reader

Is EC2 The Right Choice For Hosting Our Web Application?

I am the founder of a start-up software as a service company (applying for Y Combinator winter) and really considering using Amazon Web Services, specifically Elastic Cloud to host our PHP+MySQL application via CentOS Linux boxes. I am wondering if anybody has any input using EC2 with a production application?

The positives with EC2 are flexibility, low cost, (no hardware cost, and no co-location costs), expandability, and supposed enterprise uptime.

I have heard though in the past that EC2 has encountered downtime and various problems though.

In a perfect world, I would rather have physical access to our servers and prefer buying the hardware, but it is just not financially possible at this stage. Basically, would it be acceptable to use EC2 with say up to 500 customers?

Thanks for the input and comments.

-- Justin

View on HN

With only 500 customers, you don't need the power of EC2. There are plenty of places where you can get a Xen instance or buy a dedicated box for a monthly fee that track closely enough with EC2's pricing. What those alternatives lack is the ability to add instances (potentially dozens) with minimal notice.

I've had great uptime with EC2, but it just seems overkill for 500 customers. EC2's advantages come when you get big to the point that you need to increase capacity at a moment's notice.

Probably not. EC2 is really handy for spikes in usage but it sounds like that's not a concern for you.

Maybe something like SliceHost would be a good idea for you. Same root access, lower prices but no automated expandability.

As for "supposed enterprise uptime" -- AWS is good stuff, but it measures outages in days per year. I'm not a fan of the word "enterprise", but I don't think it allows for AWS's typical 3-8h outages.

Not knowing how much serverload 500 customers represents, it's hard to give an accurate answer.

Overall I haven't been very impressed with EC2, it still seems very experimental to me. For all of the hype about Amazons infrastructure EC2 has had absolutely embarrassing downtimes and outages.

IMO, people need to concentrate on the "E" in EC2, it's elastic and can snap back and bite you in the ass if you're not careful ;)

As a SaaS company, your uptime IS your company. This is doubly true during the early stages when you are trying to get customers to trust you and continue using your service. You HAVE to spend on your architecture, that doesn't have to mean your own private server farm, but you should get your initial webhosting from a company where that product is their primary business, not an experiment.

Thanks guys.

Seems like a standard host might be a better solution than EC2. I have been looking at SliceHost. Any other recommendations?

I was thinking of buying two slices upfront. One since running the MySQL5 backend database and the other running Apache, PHP, SendMail, DNS.

I recently looked at the bandwidth costs of S3 vs. traditional hosting, and found them less than compelling; in particular, they seemed much higher than those of shared hosting providers. EC2 has the same bandwidth pricing structure as S3, so I bet you can do better, especially since 500 isn't that large of a number.

My notes on the subject:

  http://www.mlsite.net/blog/?p=3

I've looked pretty closely at EC2 and ended up deciding against it. With any of these services (EC2, AppEngine) you're basically building your system on one big fat single point of failure.

I would rather build up an assortment of disk images pre-configured for the needs of my system's components and then deploy them as needed onto a system that supports them (eg. Slicehost).

Reply on news.ycombinator.com