Just What the World Needs: Two More Tiers of Storage
This year at Re:Invent, Amazon announced two more tiers of object storage. Now they have: six!
- S3 Standard
- S3 Intelligent-Tiering
- S3 Standard-IA
- S3 One Zone-IA
- S3 Glacier
- S3 Glacier Deep Archive
Add to this decisions about what regions you choose to use (different pricing depending on the location), the features you use, the number of API calls, and of course the much-despised egress fees, and you have a bewildering matrix of storage tier options. Thank you, Amazon, for even more choice, but are more choices what we really need?
At last year’s Gartner Conference, a storage analyst told the audience that if they are going to be moving their storage to the cloud, they should be prepared to have a full-time headcount dedicated to determining what data goes in what tier of storage. They suggested a $250,000 per year budget for this position.
To underscore how complicated everything has become with cloud storage, a cottage industry of consultants has sprung up promising to save customers money by optimizing the use of different tiers. In fact, you can now get a PhD in Cloud Economics from institutions like the University of Bristol.
In my opinion, more choice may seem like a good idea because in theory you can buy exactly what you think you need, but in fact too many choices leads to paralysis because you don’t want to end up making the wrong choices.
The ‘Jam Experiment’
The famous “Jam Experiment” shows the danger in too many choices.
In 2000, Psychologists Iyengar, Jiang and Huberman hypothesized that we are more motivated when we see lots of options because we believe that our chances of finding the right one for us are increased. They set up two tasting tables for exotic jams at a grocery store in California. One table had 24 different jams, the other only 6.
As the saying goes, the difference between theory and practice, is in theory, they’re the same, but in practice, they aren’t.
What they found was that the display with 24 jams was approached 150% more frequently than the display with only six jams. On the surface, that seems that more choices do in fact increase sales.
However, when they considered the number of purchases made they discovered that 30% of people purchased from the ‘six-jam display’ while only 3% purchased from the ’24-jam’ display.
Too many choices creates paralysis. Full stop.
S3 users I’ve talked to are always worrying about whether they are storing their data in the “correct” tier. Of course the problem with that is those choices are completely independent of the data itself, and the value of that data changes over time.
In practice, you are either overpaying or underpaying to store your data when you have too many choices.
A partial antidote to this challenge is what AWS calls their new “intelligent tiering” tier, which tries to move data to the “most cost-effective tier” based on when it was last accessed. At best, you get a delayed financial savings compared to the more expensive tier your data started life on, but it probably still isn’t truly “the right tier.”
There’s a better way to address this.
Why Not Have “One-size-fits-all” Cloud Storage?
It’s pretty well known in the industry that most of Amazon’s storage tiers are actually the same product under the hood. Just as car companies like GM and Ford build a many brands of cars ranging from luxury to economy on the same chassis, with cloud storage the data for all these tiers ends up on the same disks.
It’s a lot more profitable for GM to sell a Cadillac than to sell a Chevy, so it has to protect the high margin products by making sure that the Chevy doesn’t compete on all the “extras.” But GM has had to eliminate a number of major brands over the last 25 years because the distinctions grew too fine and customers, instead of feeling like they could choose what’s just right for them, ended up going elsewhere because they were too afraid of buying the wrong thing.
Wasabi’s approach is different. We have one class of storage.
It happens to be cheaper than Amazon’s cheapest storage, Glacier, and faster than their fastest storage, S3 Standard. So there’s no need for tiers.
We believe that cloud storage should be a commodity like electricity or bandwidth.
With the exception of edge cases like super-fast (and super-expensive) storage, one-size should fit all. And this isn’t a matter of compromising – it’s about where you should be putting your focus, and frankly, where we should be putting ours.
The benefits for the user are enormous.
You don’t need yet more software to help you figure out optimum tiering strategies. You don’t need to hire that guy that Gartner says you’re going to need. You don’t need consultants. You don’t need to write your software so that it is aware of storage tiers.
You just plug in to Wasabi’s cloud and start storing.
Tiers are Bad Enough – What Else is Blowing Up Your Bill?
Another aspect of Amazon’s complexity is all the extra charges that appear on your bill. Yes, you are charged for the amount of storage you use, but you also have egress/bandwidth fees for any data you take out of Amazon’s walled garden. And you have charges for http operations like PUT, GET, LIST, etc.
There at 39 separate line items on Amazon’s “Simple Monthly S3 Calculator.”
Wasabi has one.
No wonder people get surprised by their Amazon bill. That’s why there are so many consultants around to help you figure out what you’re likely to be spending.
Our Focus Simplifies Your Life
As an entrepreneur, I like simple. I like to spend my time thinking about the bigger issue of what my business is about and why people should care.
If I’m buried in detail, I don’t have time to think about the big picture and really add value to my organization, and to you, our customer.
I like services that make my life simple, whether it’s ordering goods online, choosing a vehicle (or just using Lyft), an all-you-eat cell phone plan, etc.
In my last company, Carbonite, we introduced the idea of unlimited backup for a flat price because we found that people hated not knowing how much their backups were going to cost.
Nobody seems to know what their Amazon bill is going to be month-to-month, and I can tell you that your CFO does not like radically unpredictable costs. It’s bad business.
While IT organizations know roughly how much data they have, generally nobody has a clue how often they access it or how many disk operations they use.
Simple, predictable pricing will always win. Nobody likes to run over budget.
No tiers = no tears.