AWS Responds with Price Cuts: Google vs AWS Pricing Round 2

I'll go with Occam's Razor: Amazon just responded to Google's newly announced pricing. To believe they guessed correctly or they have a "mole" is pretty far fetched.

At the Amazon, Microsoft, and Google level, I'd suspect that everyone is looking at the same costs in a spreadsheet and saying, "Well, I guess its time to drop to X since it costs Y -- and we might as well do that before the other guys".

If I were amazon, I would've prepared to do a price cut when Google announces one. Then when they announce it, just push the button. Why give up margin earlier than necessary? *

Of course waiting to do a price cut till their competitor does leaves a bad taste in some customers' mouths. But I bet they think they're far enough ahead that it doesn't matter. Would Netflix and similar big AWS customers really leave AWS over something like this?

* This strategy even has a name: "follow the follower". Very common strategy in sailing.

I've worked in competitive pricing before and have a little knowledge of how this works.

At all times, companies in competitive situations have a basic economic model they use for pricing. Individual sales reps are allowed to cut price to X level for orders above Y quantity, and if they want to violate that price floor they need executive approval.

Simultaneously, the companies have models that are dependent on competitive pricing -- that take into account some assumptions about price and how it affects market share. When a competitor moves in one direction, the executives consult this model to decide how to respond, but at least in the situation I was in, the final decision of how much to cut price was a seat-of-the-pants decision by the EVP of Sales.

By "model" I mean an Excel spreadsheet, but I'm sure Amazon is far more sophisticated.

Kind of frustrating, really. I just bought 2 reserved instances about 6 weeks ago. Would have saved a nice penny if I waited.

It's definitely nice to see the pricing become more affordable over time though. Before this cut I really struggled going with AWS, but wouldn't even need to think about choosing them after this reduction.

I've been wondering about this too. I'd love to see a price breakdown comparing reserved prices vs. prices with expected price drops.

There is still utility in knowing exactly what you're going to be paying, so for some people, that might be more desirable than getting the best deal.

Amazon’s Elastic Load Balancer (ELB) is not designed to handle sudden spikes. Customers need to request AWS to pre-warm the ELB to get it ready to handle the expected load. This requires a subscription to AWS support. GCE’s Load Balancer doesn’t require pre-warming. It scales instantly to support spiky traffic. Compute Engine Load Balancing hits 1 million+ requests per second with no pre-warming.

I work for a large enterprise using Google's cloud. It's quite a different experience from using our own servers to handle the same thing. We are now at the whim of a company that doesn't always announce changes until they've happened, and those changes don't always work within our business model.

Yes, it's a nightmare.

Nothing truly prevents them, but competition provides an incentive for doing otherwise.

Of course, that only works if you use provider-agnostic services (e.g. raw VMs, services with standard APIs) instead of locking yourself in to some proprietary service/API, otherwise the switching costs might be prohibitive.

Competition?

Simple Answer: Competition.

The cost per performance ratio is always going down too, which translates into lowered costs for them and ultimately the customer. They also can't collude together (at least not legally) to raise prices.

Nothing really, although on the flip side if you've got machines in a co-location data center that lease comes up for renewal and your prices are going to change. Often they will go up because the data center folks figure its harder for you to move than negotiate. As long as AWS and Google prices are 10x the cost of self running for large deployments that won't be an issue, but when they start getting closer to parity it is going to put the data center guys in a world of hurt.

It's unlikely that AWS prices would increase because they're already making some healthy profits, and there is (as others have said) more and more competition. For the competition, they don't offer the same services at AWS, and don't have the same market share, so they have to offer at a discount.

That said, there can be quite a bit of customer lock-in associated with the "premium" services, such as DBaaS, or DNSaaS which make it really difficult to switch off of a platform. That could be the hook at some point in the future to keep people paying the "AWS tax".

> The cloud provider can raise or lower prices as they see fit. What prevents these for-profit companies to keep lowering prices and not raising them?

This is different from non-cloud based providers how?

This is the old rent vs. buy debate, but for hardware instead of real estate.

Renting an apartment places a rough upper bound on your monthly expenses, since in most places, the landlord is required to do maintenance and cannot raise your rent until the lease is up for renewal.

Buying a house a rough lower bound, since your mortgage can increase unexpectedly, and you are responsible for all repairs and maintenance yourself.

So, you can never know all costs with certainty ahead of time, but if you care about predictability, it's probably better to rent (ie, go with the cloud) than to buy, because it's easier to bound your costs over a given duration.

I think the only reason prices would ever go UP, would be if the cost of generating electricity went up.. In which case, your in-house computing would cost more, as well.

Short of that, the trend seems to be costs going DOWN. Amazon recently noted their 42nd price decrease.

Heroku runs on AWS (as do other competing PaaS offerings). They'll probably pass savings on to their customers, but they have to cover costs.