Defining the Cloud: It’s a Matter of Perspective

March 15th, 2010

This post was written by Jon Toor

Arthur Cole of IT Business Edge spoke with me last week for this article. He had three questions for me.

  1. What is the cloud?
  2. Where do other types of managed service end and the cloud begin?
  3. What’s next?

Straightforward questions, but also ones that could reasonably be debated at some length. Since the objective was an column-sized piece, I condensed my response to the virtual I/O perspective.

Arthur observed this is a reasonable approach to the topic. After all, virtual I/O not only enables the cloud, it may also provide a useful way of differentiating one type of service from another. Take a look at the article and see if you agree.

 
Tags:
 
Bookmark and Share

Walking the Open Walk

March 10th, 2010

This post was written by Jon Toor

Does your vendor offer a truly “open” solution? Or do they talk about “open” while they offer solutions that are actually closed on multiple fronts?

How do you know which solutions are truly open? It can be tough to discern because there are so many different kinds of interoperability to consider.

To explore the topic fully could probably fill a book. But let me offer a simple definition:

“Open”: A solution that works with the software, hardware, and processes you have now, and with those you might have in the future.

In other words, it addresses an existing problem without creating new ones, either now or down the road.

Seems like a straightforward definition. Let’s see how Xsigo stacks up:

Does it works with my software platform?

Xsigo now works with seven operating systems and hypervisors: VMware ESX 3.5 and 4.0, Microsoft Windows Server 2003 and 2008, Microsoft Hyper-V, Oracle VM, Solaris X86, Red Hat Linux (RHEL 4 & RHEL 5). And with our recent Citrix Ready certification announcement, it’s certified for Citrix XenServer as well. More interoperability is coming, such as an upcoming release that supports KVM. Our objective is to support what our customers need, now and down the road. Read More

Bookmark and Share

Building the Open Cloud: Part 2

March 9th, 2010

This post was written by Jon Toor

In the last post, we defined the cloud. This post delves into cloud construction at a conceptual level.

It’s important to talk about the structure conceptually before diving into the nuts and bolts. The reason is that your cloud will evolve over time, in terms of the immediate objectives, the technologies you use, and your expectations. Looking at the concepts now will make it easy to see which technologies go where. And you can see how each element lays the foundation for your long term vision — while also delivering immediate value.

We can think of the cloud in three layers. Each layer offers unique functionality, and each builds towards the five defining cloud attributes discussed in the last post. But as with any layered model, each layer relies on the foundation created underneath.

Flexible architecture: The most fundamental layer is the flexible architecture. But this is really vague — what does “flexible” mean? One definition might be, “an infrastructure that can be adapted in real time to changing requirements.” But this needs more specificity to avoid a classic vendor-speak trap; if an architecture can be changed, but it takes multiple hours (or multiple people, or involves a trip to the data center), is that considered flexible?

Read More

 
Tags:
 
Bookmark and Share

Building the Open Cloud: Part 1

March 3rd, 2010

This post was written by Jon Toor

Have you noticed that when vendors talk about the cloud, the conversation often revolves around their own product or service? You may come to believe that the “cloud” is defined by a specific set of equipment and software.

It’s not. The cloud is a concept, and it can be attained via various routes and with multiple vendor’s gear. Your options are open, which is the point of this series.

We’re going to discuss what the cloud is, what technologies are involved, and how you can get there with the vendors of your choice.

Why this matters

It’s an important discussion because cloud concepts will save you time and money. For many users, these concepts already are. Every week, we meet with users who have implemented substantial portions of the cloud model. They get more done on fewer servers, and their IT people can implement projects in hours or days, rather than weeks or months.

Furthermore, they’re building the foundation for what’s to come.

Which is the second point: This is a journey. The technology does not yet support the entire vision — despite the hype! Even for those who are furthest along, there is still tremendous opportunity to do more.

What is the Cloud, anyway?

We’ll start this series what a look at what the cloud is, and get to implementation in future posts.

There are a number of cloud definitions, and then some sub-definitions (public vs private cloud, for example), but we need a definition to build a discussion around. So let’s look at two levels. Read More

Bookmark and Share

Top 10 Reasons You Won’t Like Virtual I/O

February 23rd, 2010

This post was written by Lloyd Carney

While most people like the efficiency of virtual I/O, we sometimes run into people who don’t agree. Although virtual I/O lets them do things faster, with less disruption, and at less cost, for some reason they just don’t like it. Here are the top 10 reasons why they don’t like virtual I/O, and why you may not like it either!

1) If you like spending more time at work, less time with your family : If you’re a server admin whose life is your work and you enjoy working nights and weekends, you may not like the way virtual I/O lets you manage server I/O during normal work hours on live servers.

2) No more “server is going down” messages: If you like the abuse you get when you send out those  “the server is going down” messages, you won’t like virtual I/O.  With virtual I/O, workloads can quickly be moved from one server to another, eliminating many of those extended maintenance breaks. So there’s less downtime… and less abuse from the user community.

3) Fewer billable resources: If you’re a service provider who gets paid by the rack-inch (or gets paid for network ports, or for network bandwidth consumed), you may not like virtual I/O’s increased efficiency. Your customers will use smaller servers, consume 70% fewer network ports, and keep all server-to-server traffic off the network. All of which is hard on your bottom line.

4) Fewer kilowatt hours to sell: Power companies talk green, but let’s face it – they have revenue targets, too. It probably kills them to see a Cisco Catalyst 6509 powered down. That’s thousands of watts of power consumption off the grid. And what power company exec wants to see that? Read More

Bookmark and Share

A View to the Future

February 19th, 2010

This post was written by Jon Toor

VMware last week unveiled the VMware Express, a rolling demo platform largely dedicated to VMware View, their virtual desktop environment. Essentially a powerful data center-on-wheels, the truck will provide a demonstration environment at events across North America during its year-long tour.

busXsigo was selected to provide virtual I/O on the VMware Express for the obvious reason: virtual I/O significantly increases compute density. When space and power are at a premium, you need virtual I/O to make the most of them.

Jon Owings of Veristor captured in a recent blog post several points about why Xsigo virtual I/O and View are so complementary.

Starting with a VMware View reference architecture (shown below), he figured out how it would be implemented with traditional I/O. Jon comments, “It really struck me how many I/O connections (Network or Storage) it would take to run this POD. Minimum (in my opinion) would be 6 cables per host with ten 8 host clusters that is 480 cables!”

Next, he did the same math with Xsigo and arrived at 160 cables, a 66% reduction.

ViewPOD5k

To further his point, he calculated the total server I/O bandwidth available with traditional I/O: 960Gbps was the total.

For Xsigo the total bandwidth came to 3,200 Gbps, over 3X more!

Jon surmises that a cost reduction would result from that reduced complexity (he’s right), and concludes, “Being in the data center I like any solution that makes provisioning servers easier, takes less cabling, and gives me unbelievable bandwidth. So just in the way VMware changed the way we think about the data center, virtual IO will once again change how we deal with our deployments.”

If you have chance to see the VMware Express as it tours the country, check it out. You’ll see not only the future of desktops, but of enterprise data centers as well.

 
Tags:
 
Bookmark and Share

Is the “Private Cloud” Just More of the Same?

February 12th, 2010

This post was written by Jon Toor

cloud_smThe concept of the Private Cloud sounds cool — flexible pools of resources all dynamically allocated when and where they are needed (this is also the topic of an upcoming Xsigo webcast).

But it raises an obvious question: How is this different from server virtualization? Virtualization already lets you run any application on any server. Isn’t “cloud” just more of the same?

Stephen Herrod, CTO of VMware, addressed that question during his keynote at this week’s VMware Partner Exchange in Las Vegas.

In his talk, he described cloud as an evolutionary step beyond virtualization, but a significant step, and one that needs its own definitions.

Why? Because, as he put it, “cloud is a new way of doing business.” It goes beyond just the compute infrastructure. It defines how IT delivers value to the organization. The fundamental point is this: With “cloud”, IT is built around a service model. Users select the services they need, which are then rapidly provisioned and delivered at the service levels they select. Users also “pay for” the services, although this does not necessarily mean submitting a bill each month. It could be simply providing visibility to the resources consumed.

This model has a significant benefit to IT, because it shifts the perception of IT spending. In today’s model, the CFO may wonder why IT is “spending dollars that don’t help me.” With cloud, IT expenditures are directly driven by business requirements. Read More

 
Tags:
 
Bookmark and Share

Virtual I/O Can Beat Load Balancers for Server Failover

February 5th, 2010

This post was written by Ariel Cohen

Can virtual I/O eliminate an entire hardware layer from a data center architecture to save cost and simultaneously increase performance? An IT manager I met with last week thinks so.

This New York-based IT manager is evaluating virtual I/O to solve a server failover problem – an issue for which I/O virtualization does not seem particularly relevant. However, it turns out that for him virtual I/O can indeed provide a dramatically better solution.

Here’s the basic problem. His datacenter has a number of servers that process transactions submitted by groups of clients. Different servers handle requests from different client groups.

When a server fails, clients must be re-assigned to other servers. They must either be moved to a spare server, or moved to another active server.  Either way, the move must be fast and must remain transparent to the clients. Read More

Bookmark and Share

Fortify Software Deploys Xsigo to Support Development and SaaS

February 2nd, 2010

This post was written by Jon Toor

Bookmark and Share

The Savings of Network Consolidation

January 27th, 2010

This post was written by Jon Toor

I met last week with Joe Skorupa of Gartner, and he quoted an interesting statistic regarding SAN utilization. The figures he’d seen indicated that a typical Fibre Channel port is 10X to 100X over-provisioned. That is, users are achieving 1% to 10% utilization on those ports. While Joe said the figures shocked him in their magnitude, what’s even more shocking is the financial implication of this waste.

Let’s look at a few numbers. A Director-class Fibre Channel port costs about $2,000 (with the software license, an SFP, and a year of support). A dual port HBA is about $1500. The cable adds another $100. That totals $3600 per FC connection (assuming one dual port card per connection, leaving the second port as a backup).

Since a typical FC-attached enterprise server will have two such connections, that’s $7200 per server. Which means the cost of the resources lying fallow amount to anywhere from $6480 to $7128 per server, when you consider the over-provisioning numbers. Read More

Bookmark and Share

« Older Entries