blogs.smugmug.com — Michael Arrington joked that our claim of saving $500K per year in BusinessWeek's cover story this week might be "complete bullsh*t" at the Web 2.0 Summit. The article doesn't have any other details, and I've never backed that number up - until now. See how much your startup could save!
Nov 10, 2006 View in Crawl 4
joshoNov 10, 2006
I've used it, and it's not a holy grail. It works fantastically for small setups and those with excessive storage requirements. It is not, however, good for excessive bandwidth.The costs for pushing data are very cheap up until you start in the TB/month range, and then costs basically skyrocket when compared against one (or multiple) unmetered 100Mbps boxes.For small/medium sized sites, it's badass. For online storage, it's pretty great. For enterprise and above levels of bandwidth consumption, you're going to be extremely out of pocket.One good thing is that you can get free transfer to/fro S3 accounts and EC2 virtual boxes.
jack9Nov 10, 2006
We use S3 and EC2. "but they lock you into proprietary access method " is silly, since the EC2 nodes are whatever OS template you choose. We use Fedora Core and access via SSH/SCP. We store data on S3 (free transfer from EC2s). I dont see anything wrong with parallel computing, where the money is. The S3 is just a niche extra that wasn't necesarry but will be used widely.
eskayNov 10, 2006
I think nonsecu was referring to other, non-amazon, free services with that.
bentleyNov 10, 2006
If you want to start using S3, but don't have the patience to write an app using the Amazon's API, just go download JungleDisk: <a class="user" href="http://www.jungledisk.com/">http://www.jungledisk.com/</a> . Clients for Win, OSX & Linux. IIRC, you can also use jungledisk as a mounted partition in Linux. cp /var/importantstuff /mnt/s3
jahmanNov 11, 2006
good thing they didnt take your advice :P
joshoNov 11, 2006
"I've read somewhere jungledisk's implementation is proprietary in itself. Only it can access your data."Jungledisk was my first exposure to using S3 to test drive the service before writing my own classes to implement it in some projects.It stores it to Amazon like anything else, definitely nothing proprietary about it, and you can access the stored data with anything else that knows how to talk to S3 without problem.I think it Jungledisk even has a console that shows what it's sending to Amazon (the raw http requests). The confusion over the 'proprietary' notion probably comes from the stored data being 'private' instead of 'public' by default (ie, not accessible by a web browser). Otherwise, sure, using Jungledisk you can mount S3 buckets like external hard drives and have all/most of the same functionality.