Cloud hosting is wickedly profitable(sdtimes.com)
sdtimes.com
Cloud hosting is wickedly profitable
http://www.sdtimes.com/blog/post/2010/05/11/At-last-the-truth-about-cloud-hosting.aspx
OpSource CEO lets the cat out of the bag: the cloud hosting business is waaaay more profitable than traditional hosting.
6 comments
"We've all heard the story of the nightmare application that calls the database 10 millions times an hour to add 2+2."
Actually, I've never heard this story.
This article takes many things for granted. Yes, a 1 terrabyte drive costs $100 and a cloud customer pays $1800/year for that storage - but its not like I can just buy the drive, sit it on my desk and get the same experience.
Actually, I've never heard this story.
This article takes many things for granted. Yes, a 1 terrabyte drive costs $100 and a cloud customer pays $1800/year for that storage - but its not like I can just buy the drive, sit it on my desk and get the same experience.
indeed, considering the physical security alone (or lack thereof in this case) would have had your $100 1 TB drive walk off your desk soon after you put it there.
It's pretty simple arithmetic.
Normal hosting means the customer has to get massively oversized hwardware to be able to handle traffic spikes, and then it sits around doing nothing 90% of the time. For the hoster the cost of supporting idle or fully loaded hardware is nearly the same, but the customer only gets 10% of the utility and isn't willing to pay the full price for that.
Let's say the hardware costs the hoster $50/month to buy, run and support, and yields $1000/month of utility to the customer when running at full speed. Running at 10% load, it yields only $100 of utility, so the hoster can ask for perhaps $60/month in fees.
With cloud computing, the picture looks quite different: The same $50/month hardware can now support 10 customers, and even if fees are cut by two thirds to $20 per customer, the hoster now earns $200 revenue on the same hardware, his profit margin shooting up from 20% to 300%
Of course, these numbers are completely made up and in reality cloud computing incurs some overhead and also requires some amount of idle reserves, but the basic idea holds: monetizing the potential of idle hardware yields huge efficiency gains, most of which the provider can keep for himself.
Normal hosting means the customer has to get massively oversized hwardware to be able to handle traffic spikes, and then it sits around doing nothing 90% of the time. For the hoster the cost of supporting idle or fully loaded hardware is nearly the same, but the customer only gets 10% of the utility and isn't willing to pay the full price for that.
Let's say the hardware costs the hoster $50/month to buy, run and support, and yields $1000/month of utility to the customer when running at full speed. Running at 10% load, it yields only $100 of utility, so the hoster can ask for perhaps $60/month in fees.
With cloud computing, the picture looks quite different: The same $50/month hardware can now support 10 customers, and even if fees are cut by two thirds to $20 per customer, the hoster now earns $200 revenue on the same hardware, his profit margin shooting up from 20% to 300%
Of course, these numbers are completely made up and in reality cloud computing incurs some overhead and also requires some amount of idle reserves, but the basic idea holds: monetizing the potential of idle hardware yields huge efficiency gains, most of which the provider can keep for himself.
According to James Hamilton, hardware cost is not the major factor. It's running the datacenters.
This talk by J.H. is well worth the 1 hour:
http://live.visitmix.com/MIX10/Sessions/EX01#
This talk by J.H. is well worth the 1 hour:
http://live.visitmix.com/MIX10/Sessions/EX01#
"1TB of disk space at Fry's is $100. From NetApp, it's around $700. From cloud, it's about $1800 per year. Any ops manager can do the math for ROI on that equation."
Oh God, this is how accountants look at IT. As if Amazon buys 1TB drives from Fry...
Oh God, this is how accountants look at IT. As if Amazon buys 1TB drives from Fry...
Of course they don't. There's still profit, even if you buy a big NetApp array, is the point.
Obviously there's more involved than just buying the disk. But if you already have a core competency in hosting, cloud is even more profitable than traditional models.
An article with absurd back of the envelope calculations...