You are currently browsing the tag archive for the ‘Cloud computing’ tag.
Finance & Commerce magazine published an article covering my recent keynote at the Cloud VISION 2009 event in Minneapolis. I think the article does a nice job of presenting both the benefits and risks of cloud computing. And the writer correctly articulated many of the characteristics of the new ReliaCloud platform my team is working on. I was pleased to see quotes from great local technologists such as George Reese and Tom Kieffer. Well done!
Everybody and their grandmother has heard about the Sidekick fiasco over the past month. Apparently the storage servers for the Sidekick mobile service had a bad day and Microsoft had trouble restoring customer data. The whole episode became a lead-in for a series of stories detailing potential disasters in the cloud. You see, if Sidekick has a bad week then maybe this whole cloud computing thingy is dangerous.
The media hyped up the notion that the Sidekick service represents cloud computing. Let me get this straight. So any online service that stores customer data is now considered cloud computing? This hype makes me want to drink heavily and find the keys to the nearest backhoe. Are interesting news articles such a rarity these days that journalists have resorted to this level of desperation? We stored customer data on the Internet a generation ago. It wasn’t cloud computing back then and it isn’t cloud computing now. So what is it? It’s (drum roll please…) just simply storing and accessing data via the Internet. Yeah, kind of boring isn’t it? I guess that level of banality doesn’t generate clicks and advertising dollars.
I was fortunate enough to be a keynote presenter at the Cloud VISION 2009 event in Minneapolis this past Wednesday. My presentation, “Getting Ready for the VISI Cloud”, was focused on the new ReliaCloud service platform — and specifically Cloud Servers, the first product to be launched on this new platform.
We had a couple hundred people attend the morning seminars. That’s a pretty amazing turnout on a cold and rainy October morning. It is obvious to me that cloud computing services have gained traction in the hearts and minds of IT people across the Twin Cities region.
Chris Howard from the Burton Group gave the first keynote speech. He focused on cloud computing from a very strategic standpoint — explaining how the cloud was reshaping IT within enterprises. I thought Chris gave one of the best cloud computing presentations I have heard in the past year. His presentation was clear and insightful. He skillfully blended cloud computing and socioeconomic trends. Well done Chris!
I’ve spent the past nine months working on ReliaCloud and I’m excited to see all the hard come together in a very solid service offering. Of course none of this would have been possible without a great team of application developers, system and network admins, and product managers. VISI has never launched a product with this level of automation, self-service capability, and scalability. It represents the evolution of our business and the way customers buy infrastructure.
You can check out a great synopsis of the event at the Minnov8 website. You can also find the presentations online at the VISI Blog. Here are YouTube videos of the presentations from the event: videos.
Thank you to the hundreds of people that attended the event. I really enjoyed talking with you and hearing your feedback. VISI is still accepting beta testers for the new ReliaCloud platform. The beta testing will start in early November — probably the second week of the month.
Apparently officials in the UK discovered that one of their supercomputers used to model climate change is one of the worst contributors to “pollution” in the UK. First, you have to consider CO2 to be a pollutant in order to see the irony here. Second, environmentalist are aiming their sights on datacenters and cloud providers. Here’s the deal. Say we rolled back the clock and used people, lead pencils, paper, and slide rulers to perform these calculations. Also assume we have to feed, house, and transport these people. My guess is that the supercomputer and datacenter would come out far ahead in terms of CO2 contribution. That’s just a guess.
Here’s an interesting article from a research group in Australia that performed a 7-month stress test of the Amazon, Google, and Microsoft cloud. Not surprisingly they found differing levels of service and capabilities between these cloud providers.
The synopsis: cloud vendors performance can vary by a factor of twenty depending on workloads and time of day. None of the clouds today have real SLA’s or QOS (quality of service) characteristics built into the infrastructure. It’s the wild west in the clouds right now.
BTW, I found this article through a Linkedin group called the Cloud Hosting & Service Providers Forum. It’s becoming a pretty good resource for learning about innovation in the cloud.
Jon Greaves from Carpathia Hosting described an interesting hybrid-cloud analogy relating hosting service offerings to the auto industry. I heard Jon present this analogy during a webinar sponsored by storage cloud vendor Parascale this week.
- Colocation = purchase a car, maintain a car yourself
- Managed services = lease a car, lease holder maintains the car
- Cloud computing = rent a car by the hour/day
You could also say that colocation is like owning a condo, managed services like renting an apartment, and cloud computing like a hotel room. And hopefully that hotel room isn’t like those rooms that are rented by-the-hour. 
Reuven Cohen writes about the roles of the CTO and CIO with respect to the growth of cloud computing in a recent blog posting. A CTO develops new technology and a CIO implements technology. So in a world that embraces cloud computing is the CTO role necessary?
My sense is yes. The CTO should be evangelizing and socializing cloud computing within the organization. She should discuss cloud computing strategy with customers and business partners. The CIO should operationalize cloud computing and direct the migration of services from legacy infrastructure.
A successful cloud computing architecture requires strategy, planning, and execution.
Some vendors, like VMware, will tell you it’s really easy to build a cloud. Just grab a couple servers, install hypervisors, slice off a storage LUN and you are off to the races. What was called a virtualized infrastructure yesterday is now called a cloud. I guess clouds sell more licenses. It just ain’t that easy folks.
Vendors are confused. They think cloud computing is a technology. It isn’t. It’s a business methodology that leverages technology and business process. You can’t buy it in a box. Trust me; I’ve looked at a lot of boxes lately.
So how do you go about building a cloud computing platform? Well a cloud has four basic components:
The virtualization layer, data storage, and computing components are fairly straightforward. You can buy most of these commodity pieces off the shelf. The management system is the most difficult component to build or purchase. Let’s look at these cloud components in more detail.
It seems like virtualization and clouds are almost synonymous these days. Virtualization is a technology and cloud computing is a business methodology. I’m going to keep hammering that point home so bear with me. The reason these terms are confused is because pretty much every cloud computing platform uses some form of virtualization technology. Can you build a cloud without using virtualization? I guess. Heck service providers are calling almost anything they sell a cloud these days. In reality it’s hard to build a cloud without using some form of virtualization. One of the main features of a cloud is infrastructure abstraction — or the virtualization of the infrastructure from the perspective of the cloud user. Virtualization technologies are simply the easiest, cheapest, and most efficient way to provide this type of abstraction layer.
So which virtualization technology should you use to build a cloud? The first thing to understand is that from the perspective of the user the virtualization layer does not matter. It really doesn’t. Cloud users could care less whether your cloud employs VMware, Xen, or Hyper-V virtualization technology. The users only care about service availability and performance. So don’t pick a virtualization technology based on market share or public awareness. Ultimately the cloud virtualization technology should be based on your cloud management platform. The management platform will manage the provisioning and configuration of infrastructure within your virtualization layer. So it’s critical that the management platform and virtualization technology work together.
You could pick VMware as the virtualization layer because you plan to build a cloud management platform around vCenter and the VMWare APIs. The same goes for Hyper-V and Microsoft Systen Center Virtualization Manager. Or maybe you want to build a management platform around Xen and the Citrix C3 doohickey. The key point is that the cloud management platform really drives the virtualization technology decision. I’ll talk more about the management platform in a bit.
(To Be Continued)
I’ve often felt like a sports fan watching a great match from the sidelines as cloud computing developed over the past few years. I knew that cloud computing was changing the rules of the game for the teams in the hosting industry. And while my team was still winning games the competition was sporting sleek new jerseys. Well that is going to change.
I’m leading the charge at VISI to build the largest infrastructure cloud computing platform in Minnesota. Our cloud will provide enterprise-class features at near-commodity pricing. We will leverage new and old technology. We will support a wide range of infrastructure environments at a scale that is currently unheard of in our region. And we will launch before the end of the year.
I’ve had to answer so many questions over the past few months. What is cloud computing? Do you really want to compete with the big clouds? Why not just resell Amazon’s cloud? Which hypervisor will your cloud use? Do customers care which hypervisor you use in the cloud? Will our businesses pay for computing on an hourly basis? Each answer to these questions seems to bring up three more questions. Even though much of the cloud architectural planning is complete our work has only begun.
Let’s start at the beginning. Why is my organization building a cloud?
We are building a cloud because it is the most efficient way for a service provider to deliver hosted services. It is also the most efficient way for customers to buy hosting services. Our national competitors are launching cloud computing platforms on a weekly basis. Our customers are asking us about our future cloud strategy. We recognize that if we don’t change and adapt our hosting business model that segment of our business will eventually die. If we want to stay in the hosting business we have to invest in the cloud. It’s that simple.
The next question I hear is: why do you want to complete with big cloud providers like Amazon, Google, and Rackspace?
I don’t. I want to stay clear of the large cloud providers. I don’t want to get into cage matches with 800-lb gorillas. We’ve built our business in the gaps. Some service providers focus on the mass-market, commodity side of the business. Others focus on the largest global enterprises. We’ve been able to flourish and grow in the middle. If you are a small business with complex needs or a large business with price sensitivity then VISI is usually a good fit. I believe we can capitalize on the gaps in the clouds. We can build a cloud that competes with Amazon pricing. But we can also build a cloud that has the security, high-availability, and support of a Terremark. All clouds are trending towards the middle; so why not just start there.
One of the biggest questions I’ve had to face is the “buy vs. build” question. Should we design a cloud platform from scratch or buy a pre-packaged cloud solution and customize? The developer in me wants to build. The business guy says buy. The challenge with buying is that enterprise-class cloud computing platforms don’t exist on the market. (I’m sure some vendors would beg to differ.) Building a cloud platform from the ground up would take an experienced group of developers, some serious dough, and at least a year.
We’ve decided to use a combination of commercially developed components and VISI-developed components in our cloud platform. We will leverage both open-source and proprietary technologies. We will integrate with existing automation and support tools. The commercial components will speed up our platform delivery time while our in-house developed components will provide better service integration and differentiation in the marketplace.
Ultimately our customers will decide whether we are making the right decisions or not. I plan to give them a voice in the development and testing process. This blog is an open invitation for a front-row seat to watch the build.
I have tons more stuff to talk about. But I won’t pile everything into a single post. I’ll provide more insights on this blog in the coming weeks as my team progresses on our cloud platform design.
I was recently interviewed for an article on cloud computing in the Twin Cities Business Journal. The article specifically mentions that my company, VISI, is planning to launch a new cloud computing service. Did I mention that I was building a cloud? You didn’t think I was going to sit on the sidelines did you? More on my cloud building effort later…
