Friday, August 22, 2014

"Cloud Storage Is Eating The World Alive" Really?

I'm naturally happy when someone cites my blog and uses my data, as Alex Teu did in his post Cloud Storage Is Eating The World Alive on TechCrunch. I'm less happy with the some of the conclusions Alex drew. Below the fold, I argue with him.

Who is Alex Teu?
Alex Teu is the vice president of business development at Oxygen Cloud and odrive.
Oxygen Cloud is a service to: 
Centrally organize all your company files. Organize all your shared directories the way your teams work. Easily manage all your user groups and permissions.
It is layered on cloud storage: 
No more headaches with unlimited cloud storage. Forget about server cycles, hardware maintenance, and limited disk space. Get storage that grows with you.
Lets just say Alex has a vested interest in convincing you that the cloud magically solves all your problems. He's not exactly cautious in his rhetoric:
The cloud storage war being waged by Amazon, Google and Microsoft has been well publicized, and is resulting in the cost of all cloud infrastructure including storage racing to zero.
He has an impressive graph to support this claim, based in part on data from my post Cloud Storage Pricing History. See how the cost of cloud storage is going to zero! Just project the last part of the graph forward in time and see how soon you get there:
Where do we go from here? Undoubtedly, to zero. When? Perhaps as soon as this summer or by year end. Will it literally go to zero? Maybe no, maybe yes. If you only had to pay $.001/GB, that works out to $1/TB. Oh yeah, that feels like zero!
There are a number of points to be made here. First, even Alex admits that right now the storage price is $0.024/GB (per month, which he doesn't say), so his target of $0.001/GB/mo is a factor of 24 less than current prices. From his graph, we see that the price has dropped by a factor of 6 in 9 years. So a different projection as naive as his would predict that the $0.001/GB/mo price point would be reached in 2050. A long time to wait for "free".

Second, the price Alex quotes is for Amazon's Reduced Redundancy Storage, which is less reliable:
This durability level corresponds to an average annual expected loss of 0.01% of objects.
S3 is currently $0.03/GB/mo. So to get $0.001/GB/mo reliable storage we would have to wait until 2059.

Third, I agree with Alex that constructing a history of cloud storage prices is difficult:
The data research for the top graph involved a mix of public announcements, a search of the Internet Archive and efforts from others who attempted to track the price changes such as David Rosenthal. This also involved some guesstimation on my part for some begin and end dates of a price change as I was able to ascertain a price on a given date but was not able to confirm with a public announcement.
But irrespective of the quality of the data, the graph is misleading. By drawing lines between the data points Alex suggests that prices have decreased smoothly. They haven't. They have stayed steady for various lengths of time and then changed stepwise. Its true that recently the step changes have been frequent and large, and Alex provides the explanation:
As you can see from the graph below, Amazon was the only real game in town for several years, and the price of cloud storage hardly moved the needle. It was only after Microsoft and Google got into the game in 2010 that we saw any real price movements.
The reason why Alex's projecting of the recent price changes into the future is wrong is that the introduction of competition into a previously monopolized market is a one-time thing. There isn't a queue of other Microsofts and Googles lining up to enter the market. Yes, its likely that the market will in future be more competitive than it was when Amazon was the only game in town and their margins were obscene. But that means it will move in line with the underlying cost of storage, which will be much less free than it used to be.

Fourth, the $0.024/GB/mo and the prices on Alex's graph are just the storage rental, they are what you pay if you never access the data and you never move it out of Amazon. You need to pay up-front for local storage, but once you've done that access is effectively free.

Fifth, Silicon Valley and other place are overrun with startups building all-flash and flash-disk hybrid storage systems. Why would people pay much more for these systems than for conventional storage arrays? Because they are blazingly fast, which cloud storage isn't.

Alex correctly points out that traditional storage vendors are having a hard time, comparing Q1 2013 with Q1 2014 for 6 large vendors, with drops from 20% (IBM) to 2% (NetApp). At The Register, Chris Mellor has a more detailed discussion of this phenomenon:
If we chart the recent quarterly storage revenue numbers from the mainstream array suppliers, we can see that some mainstream storage suppliers are feeling a revenue pinch. Take, for example, Dell, HP, IBM and NetApp. HDS's revenues are not growing particularly quickly and EMC storage (EMC Information Infrastructure excluding VMware and Pivotal) revenue growth is also not that impressive.
Where Alex and Chris differ is not the revenue drop, but the cause of the revenue drop. Alex is sure the magic of cloud is the cause:
It’s well established that cloud storage is 10X better. It’s more accessible, more secure, elastic, pay-as-you-go.
Today, we are all on mobile devices, work everywhere and generate ever increasing amount of data. Our work and personal lives are set up as the perfect playground for cloud storage.
Now that the cost of cloud storage is near-zero, the business choice to abandon clunky, expensive traditional storage is a no brainer.
Chris agrees that the cloud is one factor, but puts it last in a list of 9:
  • Object storage
  • All-flash arrays
  • Hybrid flash/disk drive arrays
  • Server flash storage
  • Software-only SANs and filers
  • Virtual SANs
  • Converged systems
  • Software-defined storage
  • The cloud.
It obviously isn't "well established that cloud storage is 10X better", presumably than local storage, since there are 8 other technologies that Chris thinks are partly causing the mainstream storage array business to shrink.

Lets get real. There are many situations in which cloud storage makes great economic sense. They include when demand is very spiky, when it is growing unpredictably, or when capital is extremely expensive. They do not include base-load long-term storage. And committing all your storage to the cloud carries all sorts of risks that must be set off against any potential savings, from foreign government action to, as The Register recently reported, Total Inability To Support Usual Performance.

PS: This is my second post in a few months debunking credulous techno-optimism from the tech media. I hope it doesn't become a habit.

No comments: