Musings on cloud storage as a business.

[DrKK]

So I was just meditating on this.

At this moment in time, so, mid 2015, there are plenty of per GB services that will store your data at a price point of around $0.01 per GB per month. Amazon Glacier, for example. Naturally, that works out to $10 per TB per month. Enterprises are going to be paying (at this moment) about $40 per *raw* (i.e., not arrayed, non-backed-up) TB of HDD storage, just for the equipment. That equipment breaks, there are servers to buy, backup servers (with their own storage) to buy, there are employees to pay, there are electricity bills to pay, there is ingress and egress bandwidth to pay for, and there is constant downward pricing pressure from a whole slew of competitors in the uber-l33t cybar-cl0u|) space. At the end of the day, the way I figure, a customer has to be on board with your cloud storage for at least 8 months before the cost of even providing him with the capacity is made up, and then after that it's peanuts, and he can leave at any moment for a competitor.

Sounds like a tough business to be in. Margins can't be very good. Lots of competitors.

Probably better off buying a Subway franchise.

That is all. Just a philosophical meditation, and an invitation for thoughts on the subject.

 

[anodos]

I think the key is charging more than competitors but providing better service (support, ease of use, etc). Not exactly fertile ground for a startup.

 

[mjws00]

The price is pretty elastic, based on customer sophistication, and services desired. Consider rsync.net at 0.20 per GB to start. Now we are at 2 months for return on capital expense. That is pretty unheard of in most businesses.

Op costs and bandwidth would seem to be the scary things. You have to scale pretty hard to make moving 100's of TB cost effective.

There is also that initial seed and disaster recovery problem/opportunity. How to seed and recover 20TB remotely when a business has it's infrastructure crippled. Even over night courier isn't gonna cut it for some.

Fun one to analyze. There are definitely viable models if one specializes or localizes.
But running straight into Amazon or Google jaws might be a very poor choice.