The Virtuous Cycle of Cloud Computing

The Virtuous Cycle of Cloud Computing

  • Comments 4
  • Likes

In the Day 1 keynote at the recent re:Invent conference, there was an interesting point made about the virtuous cycle that can occur for the cloud vendor and for customers. As I listened to the keynote, I kept thinking: “They are missing the biggest benefit for the entire industry; if the public cloud vendor has the right strategy and is thinking about how to benefit the largest population possible, then they are completely missing how this virtuous cycle can grow to benefit every organization in the world – even if they are not using the public cloud.”

Let me explain a bit more about what I mean.  (And, before I get too much farther along, I want to note that this post ties into the cool news yesterday about our work with the Open Compute Project.)

The virtuous cycle of a public cloud looks a lot like the image below.  As the usage of the public cloud grows, you need more hardware to meet demand – and for sustained growth you will need a lot of hardware. This need for hardware increases your purchasing power and you can then negotiate lower prices as you purchase in bulk. As your purchasing power grows and your costs drop, you then pass those savings on to your customers by dropping your prices. The lower prices increases demand and the virtuous cycle continues.

Virtuous Cycle Complete - public

For customers using the public cloud, they can see the benefits of this virtuous cycle (the lower prices) – but what about organizations that are also using private and hosted clouds? How can they gain benefits from what is happening?

Organizations with multiple clouds can benefit if (and only if!) that public cloud vendor has at the core of its strategy an intention to take everything that it is learning from operating that public cloud and delivering it back for use in datacenters around world – not just in its own.

This is where Microsoft is so unique! Microsoft is the only organization in the world operating a globally available, at-scale public cloud that delivers back everything it is learning for use in datacenters of every customer (and, honestly, every competitor). Our view is the learning that we are getting from the public cloud should be delivered for all the world to benefit.

This innovation can be seen by applying these public cloud learnings in products like Windows Server, System Center, and the Windows Azure Pack – and these products are the only cloud offerings that are consistent across public, hosted and private clouds – ensuring customers avoid cloud lock in and, maximize workload mobility, and have the flexibility to choose the cloud that best meets their needs.

With this in mind, I want to show you how I think the virtuous cycle can and should look – and how it can benefit any organization in the world.

First, at the center of this virtuous cycle is incredible innovation. This means innovation in software, innovation in hardware, and innovation in processes. When you are ordering and deploying 100,000’s of new servers and xx bytes of storage every year – you have to innovate everywhere or you will literally buckle under demands and costs of procuring, deploying, operating, and retiring hardware at this scale.

Microsoft is addressing this challenge in the most direct and complete way possible: Over the last three years, Microsoft has spent more than $15B building datacenters around the world and filling them with the hardware and capacity demanded by customers of Windows Azure and other Microsoft cloud services.

We keep our public cloud costs low by managing our supply chain for this kind of capacity, and, per the cycle, we pass these savings to you. We also carefully track things like the number of days from when we place an order for hardware to the time the order appears on our docks (“order-to-dock”), and then we track the number of hours/days from “dock-to-live” where we literally have customers’ workloads being hosted on that hardware. Throughout this process we set aggressive quarterly targets and we work constantly to consistently drive those numbers down. If we didn’t have a best in class product and performance, it would be impossible to remain profitable at this kind of scale.

As you can imagine, after spending $Billions on hardware every year, we are highly incented (to put it lightly) to find ways to drive our hardware costs down. The single best way we have found to do this is to use software to do things traditionally handled by hardware. For example, in Windows Azure we are able to deliver highly available, globally available storage at incredibly low prices through software innovations like SDN – all of which is based on low-cost, direct-attached storage. This brings storage economics never before seen in the industry.

One example of this is the most common workload hosted in Azure: The “Web” workload. Whether it is Azure acting as the web tier for hybrid application, or the entire workload being hosted in Azure, the web workload is a part of just about every application. This makes it a great place for innovation. In Azure we pioneered high-density web site hosting where we can literally host 5,000+ web sites on a single Windows Server OS instance. This dramatically reduces our costs, which in turn reduces your costs.

At Microsoft, we think the public cloud’s virtuous cycle can actually get a lot bigger, a lot more functional, and a lot more powerful by integrating service providers and hosted clouds.

Virtuous Cycle Complete

Not only is this expanded virtuous cycle more practical, I’m sure it also looks familiar to what is already up and running in your organization.

There are some pretty solid examples of innovation that was pioneered in Azure and then brought to the whole industry for use everywhere through Windows Server and System Center:

  • For highly available, low-cost direct attached storage, in Windows Server 2012 we shipped a set of capabilities we call Storage Spaces. Storage Spaces delivers the value of a SAN on low-cost, direct-attached storage, and it has been widely recognized as one of the most innovative new capabilities in Windows Server – and it was significantly updated in Windows Server 2012 R2.
  • Service Bus provides a messaging queue solution in the public cloud that can be used by developers for things like a queuing system across clouds and building loosely coupled applications. Check this post for an in-depth review of Service Bus. Service Bus also ships as a component of the Windows Azure Pack – providing value pioneered in the public cloud for use in private and hosted clouds.
  • Earlier I referenced the ability to host 5,000+ web sites on a single Windows Server OS instance. This has had an obvious economic impact on of costs of Windows Azure where we host millions of web sites. We proved that capability in Windows Azure, battle-hardened it, and now it ships for customers around to world to use in their datacenters as a part of what we call the Windows Azure Pack (WAP).

This is what it looks like when the complete virtuous cycle is in effect.

Our efforts haven’t been limited to software, however. Our innovative work with hardware in our datacenters has driven down costs while at the same time increasing the capacity each core and processor can support.

Our work with hardware was highlighted yesterday when we announced that we are joining the Open Compute Project and contributing the full design of the server hardware we use in Azure. We refer to this design as the “Microsoft cloud server specification.” The Microsoft cloud server specification provides the blueprints for the datacenter servers we have designed to deliver the world’s most diverse portfolio of cloud services at global scale. These servers are optimized for Windows Server software and can efficiently manage the enormous availability, scalability and efficiency requirements of Windows Azure, our global cloud platform.

This design spec offers dramatic improvements over traditional enterprise server designs: We have seen up to 40% server cost savings, 15% power efficiency gains, and a 50% reduction in deployment and service times. We also expect this server design to contribute to our environmental sustainability efforts by reducing network cabling by 1,100 miles and metal by 10,000 tons.

This level of contribution is unprecedented in the industry, and it hasn’t gone unnoticed by the media:

  • Wired: Microsoft Open Sources Its Internet Servers, Steps Into the Future
  • Forbes: The Worm Has Turned - Microsoft Joins The Open Compute Project

These are just a couple examples of innovation that is happening here at Microsoft – innovations in process, hardware and software.

At Microsoft, we recognize that the majority of organizations are going to use multiple clouds and will want to take advantage of Hybrid Cloud scenarios. Every organization is going to have their own unique journey to the cloud – and organizations should make decisions about cloud partners that truly enable them with the flexibility to use multiple clouds, constant innovation, and consistency across clouds.

This is an area that we focus on every day, and you can read more about it as a part of our ongoing, in-depth series, Success with Hybrid Cloud.

Your comment has been posted.   Close
Thank you, your comment requires moderation so it may take a while to appear.   Close
Leave a Comment
  • 話說這是一件很不容易的事~經過漫長的煎熬總算是有所小成,在此我個人有許多感慨有一種衝動想要訴說一下:表面上虛擬世界的帳目很好看但有誰知道我是如何撐過來的?網路上對我的一切優惠我是望城莫及從來也沒有購過物也談不上其他增購硬體設備了?一直以來我始終無法在我們國家申請一張信用卡~原因是我無財力級新資公司行號可證明只說我是網路開發人員銀行是不認可的?再說公司在台灣並無���接的對話窗口我無法取的任何可以證明我自己的一切事務! 我有很多無奈是否請大家幫忙我想想辦法~試問一個人這麼久沒有任何收入日子要如何過可以告訴我~在台灣我應該找某一家在台子公司或者某人士接觸以解決我的困難?然而我應該備查所需的什麼樣的認證文件資料`網路上我所注冊的一切都是真實都可用來與我連絡的?請你們幫助我謝謝?在此跟大家道歉實在不應該在此說這些~但我已用盡一切辦法總是不得其門而入請多包函?

  • The problem with cloud computing is you guys just want to be able to easily spy on our data. I'm an ICT enthusiast bordering on geeky and I can't believe how the world fell for this. guys check this out. I convince you to stop storing info on your computer and instead store it on the net, access it from everywhere, for peanuts or better yet, free. Then I start mining the data to pick up your habits, needs wants etc, and sell this to the ad guys, who then use it to target you with adverts psychologically designed especially for your character type. Dude, you end up buying stuff you don't even need and somehow convince yourself you need it. Then once you are hooked I start charging you. you gotta hand it to the guys who thought this up> They are brilliant.

  • Enterprises, too, are incentivized to find ways to drive hardware costs down. Take note from this large CSP, who has 'been there' already: "The single best way we have found to do this is to use software to do things traditionally handled by hardware." I couldn't agree more. Be wary of vendor captivity and hidden costs, a multi-vendor approach will keep them on their toes. Given bandwidth constraints and application performance requirements, look heavily at hybrid cloud solutions for your best chances of cloud success.