Microsoft Windows Azure IaaS Price Promise? It’s A Loss Leader

Peter Judge has been involved with tech B2B publishing in the UK for many years, working at Ziff-Davis, ZDNet, IDG and Reed. His main interests are networking security, mobility and cloud

Microsoft is matching Amazon’s IaaS prices – but what it really wants to sell is a fully Microsoft PaaS-based cloud, says Peter Judge

Microsoft’s Windows Azure will match the prices of Amazon Web Services (AWS) for commodity products like server instances and storage. But it’s not a price war – those products are a loss leader.

Infrastructure as a service (IaaS) has been available in beta on Azure since last June. Now it goes into general availability, and Microsoft wants it to stand against AWS, so has promised to always match Amazon’s prices. But what Microsoft really wants to sell you is not IaaS, but lots of very Microsoft-specific platform as a service (PaaS) cloud services.

That’s where it thinks the action will be in future.

microsoftmichaelnewberry

Some IaaS prices go up

There is a small irony in the fact that some IaaS prices have actually gone up as the service goes to general availability. A fairly typical two CPU Windows instance  has gone up two cents per hour, from $0.16 to $0.18.  But that’s because the beta IaaS service did not come with a financially-backed service level agreement (SLA) and was priced low to get people using it.

That has had some success. Microsoft’s British Azure lead, Michael Newberry (pictured), claims 1.4 million instances have been spun up on the IaaS beta service – and the standard offerings were re-sized somewhat in response to how Microsoft saw those VMs being used.

Paas features and hybrid apps

Now it’s fully launched, Newberry wants to see even more action there, but the price promise is just to simplify the customer conversation: “Please take for granted that as prices change we will match any reduction that may come out of AWS,” he tells TechWeekEurope. 

But the conversation is being simplified, so Microsoft can move on and talk about its favoured option, PaaS.  “What we’d really like to talk to you about is how our platform is differentiated,” he tells us.

In Microsoft’s view, the world is transitioning from “client and server” to “device and cloud”. In the process, new apps are being built on the cloud, while old apps are packaged up and moved there. The old apps have to go on IaaS (it offers old-style infrastructure, but virtualised) while Microsoft wants new apps to be built on the more authentically-cloudy PaaS.

PaaS is better than IaaS for developers because Microsoft can offer more powerful (but arguably more proprietary) services, says Newberry – and he is joined on the line by a partner, Dan Scarfe, who is CEO of Dot Net Solutions. “Platform a service is our recommended approach to leverage the  power you can get from cloud,” says Scarfe.

In Azure’s PaaS, apps get to use things like Microsoft Active Directory and mobile services, as well as features for specific areas like media, says Newberry. Also, he offers the “single throat to choke” argument for getting the whole thing – applications and stack – from a single vendor: “For things like SQL server, where Microsoft is the software vendor as well as provider of platform, you get support across the stack.”  Vendors can’t blame each other if it all goes wrong.

“Those kinds of value-added services are where the industry is going to evolve into as part of the whole movement to device and cloud,” says Newberry. “That’s what the cloud OS proposition is all about.”

Newberry also suggests running hybrid apps: once the legacy application is on Azure’s IaaS, a new front end can be built using PaaS. In that way the new PaaS stuff can encapsulate the older code.

Is anyone using it?

But is anyone really doing all this? Newberry is utterly mysterious on real user numbers.

It’s obvious, from AWS’ enormous business in the cloud, that users want plenty of IaaS services, and Microsoft’s PaaS game is a minority interest right now. Newberry and Scarfe put this dominance down to the oft-quoted fact that most IT effort goes in keeping the lights on.  “The typical customer willl spend more of their budget on older apps and not so much on running new stuff,” he says.

So if users do flock to Azure’s vanilla IaaS offerings, how many will be upsold to the “more functional” Azure PaaS? He won’t say.

What about the amount of users on Azure in total? Along with the IaaS price announcment, Microsoft has issued a claim that it has 200,000 Azure customers. However, it won’t say how large they are, or whether they are testing Azure or using it for real.

We don’t think many users are really putting crucial loads on Azure, for one simple reason. In February, the service suffered a catastrophic failure. The cause was a failure to distribute a current SSL security certificate to all its servers – and the result was a collapse of all Azure data centres.

The point is this: Microsoft has flagship Azure customers, like Easyjet. But when Azure failed, those customers, and all the customer facing applications we could see – continued completely unfazed. The unavoidable conclusion is that no one is actually relying on Azure for major, live apps. If people are really using Azure, why was no one hurt when it broke?

We put this to Newberry, and the silence is deafening.

He hands the question to Scarfe who eventually says: “The bad news is that it did affect some customers and there were some high-profile customers taken offline,” but he won’t say who they were.

In any case, Newberry says, he would advise customers to build in redundancy and resiliency. In the case of a global Azure outage, of course, that would involve using backup outside of Azure (which might be hard to manage if you are using Microsoft-specific Azure PaaS features).

However, the good news is there should be no need to doubt Azure’s future reliability, he says. “In our root cause analysis, we discussed why that set of circumstances shouldn’t reoccur.”

In other words, you have to take Azure on trust. Amazon’s AWS has a patchy reliability record as well, of course. But it has plenty more instance-hours of uptime to its credit than a service which is still little more than a flea-bite on the cloud giant’s back.

Microsoft’s hope is that users will prefer to use a single-vendor PaaS cloud approach than building reliability on a more open IaaS offering. That may well be true – but it will take more than a cut-price Amazon clone service to convince users to buy into the Microsoft vision.

How well do you know Microsoft? Try our quiz!