Hacker News Reader

Comment by shingen | Hacker News Reader

A monetized Posterous could be run on 100 or less servers. Tumblr runs on perhaps 1,500x according to High Scalability, and it's at least a few hundred times more nasty than Posterous in terms of server thrasing (pageviews and hits on the data via the dash).

Tumblr daily pageviews: 500m; Posterous daily pageviews: 1m

Posterous would never be a billion dollar service, but it could be the type of service that generates $10m+ in annual profit.

100 servers from a competent host + bandwidth via a pooling deal = less than $50,000 per month. It wouldn't be sexy, but it'd hold.

I wouldn't keep the current Posterous business model of chasing Tumblr into a pool of red ink. I'd place a bet that users will pay for a service that will stick around. It would reduce usage, but there's a large base to monetize that would be willing to stick around.

View on HN

Uh, what?

1m pageviews a month is about equivalent to 10/second. A single app server can handle that easily. Add on a redundant app server, two database servers, and some static/queue/mail ones, and you're looking at 10 servers _tops_ for that level of traffic (and much more).

I'm guessing he meant uniques.

OP must've miscalculated, surely. 0.1 pageviews per machine per second is quite amusing though.

"We're really struggling here cap'n":)

Well, to be honest...

Quantcast actually gives 1 million pageviews a day (and not a month), which is, yes, 10/s (and 550 times less than Tumblr).

Either Quantcast numbers are very off, either Posterous can indeed run on very few servers.

you have to expect it to peak a lot higher though.. The requests are probably not going to be distributed evenly over 24 hours..

(edit: but yeah, you only need 100 servers if you're running something with "enterprise" in the name ;)

It's obviously 30 million monthly pageviews (I said 1 million daily pageviews, I think you misread my text).

100 servers would assume media hosting for images and other files that go with a blog. That could all be kicked out to other services like Amazon, but the costs would obviously go up and I don't think it would be necessary.

At $500k +/- a year in infrastructure costs, I think Posterous could be wildly profitable. You could run a tight ship with five to ten people (throw in $500k to $1m in employee costs). If someone put a gun to your head, it could be maintained with a two or three person team.

I mistyped my original response. 1 million daily pageviews is still 10/second. The rest of my post remains the same, so 30 million monthly pageviews can easily be handled with 10 or so servers.

Reply on news.ycombinator.com