Ask HN: My site is getting popular, bandwidth expensive
What should I do? Obviously I'd like to keep running the site but not if it starts costing me a lot of money.
What should I do? Obviously I'd like to keep running the site but not if it starts costing me a lot of money.
0.10 to 0.15$ per GB is pretty standard low-end for low bandwidth sites. Prices only really drop once you cross the 50-100TB/m range.
Your problem is that you're hosting cost (linode) is something you need to pay regardless, so you want to use that "free" bandwidth. I've seen people swap domains. Have images1.mysite.com which points to their main servers...and then when they get near their limit they switch to images2.mysite.com which would point to a place with cheaper bandwidth. Problem is, linode prices are pretty damn reasonable (probably cheaper than S3 when you add storage/GETs/etc).
If you prepay at linode it's $0.10/gb...you should just buy a Linode 512 at that point, since it's the $20/200gb ($0.10/gb again) + you get the extra machine.
100tb.com, a softlayer reseller, is a known large-bandwidth [over]seller. They have VPs for ~$100/month with 10TB included..there may be other options, but I probably wouldn't recommend this route.
I hope you are making backups of your images...
The issue you're going to find is that bandwidth doesn't get much cheaper than Amazon S3 unless you start going with hosting providers who are questionable. At $5/day with Amazon can we assume you're transferring around 30GB/day?
At this point a decidated hosting provider may be a better fit. For $40 a month you can get a dedicated box with Hetzner[1] that will allow you 2TB/month @ 100Mbit which then gets shaped to unlimited @ 10Mbit. 2000GB/31 days is basically 65 GB per day - more bandwidth than you need for the moment and at a cheaper rate than S3 can provide (2TB at Amazon S3 rates ($0.15/GB) is $300). If you can survive with lag from Germany this would be optimal.
The other (far nicer) option is to try and find a strategic partner. Someone like ICanHasCheezburger / FailBlog makes enough money from eyeballs that it may be worth their time to either provide you with money or bandwidth in exchange for a major link back to their page. Reach out to them, show them your page stats and see what sort of response you can get. They're already in a far more lucrative position as far as ad partners are concerned and can likely use your traffic to move into an even better bargaining position. They likely also have huge amounts of bandwidth spare as well which they may be more than happy to offer to you.
As another commenter said, $5/day won't kill you and I think you have quite a reasonable chance of making this back. Congrats on your success so far - if you ever come down to Sydney come along to SiliconBeach - I'd be interested to chat =]
I won't kill off the site just yet, I'm prepared to lose a bit of money on it while I'm trying to figure out a solution. Thanks for the suggestions!
With thousands of visits, it's worth looking at your HTML/CSS and simplifying it to minimize the bandwidth needed. May seems small in comparison to the image bandwidth, but it won't hurt.
Very sound advice. Doing this has produced amazing results for us more than once in the past. As an added advantage, it'll speed things up a bit too! :-)
Just my thoughts.
* minify HTML output
* minify your CSS files
* use the Google hosted version of jquery http://code.google.com/apis/libraries/
Although I'm kind of reluctant to admit this, I messaged the URL to a whole lot of my contacts. :-)
The more I think about it, the more I feel it one of those typical things that have the potential to go really viral; something like the hampsterdance.com of 1999/2000.
My suggestion is to start promoting it every which way you can and building some kind of community around this asap. (In all probability, you won't have to actively do any of these for long.)
All the best.