VMworld: Cloud powered by Xsigo

September 1st, 2010

This post was written by Jon Toor

VMworld is all about the cloud. And what you’ll find in that cloud is Xsigo.

Behind the scenes at VMworld, now happening in San Francisco, there are 14 Xsigo I/O Directors connected to hundreds of physical hosts. This gear, which is running hands-on labs, booth demos, and classes, is located at several locations around the country. I’m told that one is on the east coast. The idea of course is to help validate the cloud concept. You don’t care where the server is, just that a server is always available and accessible.

Click to the next page for a look behind the scenes. Read More

 
Tags:
 
Bookmark and Share

10 Ways Ethernet-attached Virtual I/O Changes Everything

August 31st, 2010

This post was written by Jon Toor

Virtual I/O for every server on earth. Does that sound like a stretch?

It’s not. Xsigo just made a huge advance that dramatically reduces the cost of virtual I/O: it now works with standard Ethernet.

Here are 10 reasons why this changes the game.

10) Connects to the servers you have… right now: For the first time, virtual I/O will work with the server ports you already have. No need to install new cards. Just connect the server to the I/O Director via Ethernet, load the driver, and you’re ready to deploy virtual connections that will carry both Fibre Channel and Ethernet traffic, all over standard Ethernet. Get up to 64 virtual connections to a server without touching the hardware.

9) 10G and 1G flexibility: You can use either 1G or 10G Ethernet ports. Pick the I/O that meets your performance and cost needs.

8 ) 10G LOM for even greater value: The value gets even stronger in early 2011 when 10G Ethernet ships embedded on commodity servers. When that happens, this low cost 10G connectivity means you get the performance of FCoE (and greater functionality!) without the cost and hassle of a CNA.

7) Choose the highest performance or the easiest integration: Do you need more than 10G performance? Only Xsigo lets you choose from 1G, 10G, 20G and 40G connectivity (20G and 40G are InfiniBand). All can be managed from a single graphical  user interface or command line interface.

6) Commodity servers OK: Get virtual I/O on your standard server configuration. Regardless of the vendor or the server type (rack or blade), there’s no need to specify unique configurations for virtual I/O.

5) Evergreen technology: You can always add new I/O types by simply adding modules to the I/O Director. Want to connect to FCoE-native storage down the road? Just add a module. New I/O types can be made accessible to all attached servers without any re-cabling.

4) Use the cabling you have: You may be able to re-use your existing cabling for even easier integration. Only Xsigo gives you that option.

3)Proven design: Xsigo’s Ethernet-platform is identical to the InfiniBand-based platform. Same proven OS (now at 3.X level) and the same proven silicon. So all the maturity of the InfiniBand product translates directly into the Ethernet product.

2) Under $500 per server: Complete solution deployment costs can drop below $500 per server for 1G Ethernet server connections. This gives you new flexibility to use virtual I/O in cost-sensitive applications.

1) No CNAs: Why buy an FCoE converged network adapter when you can use the ports you already have? CNAs may cost you thousands of dollars per server. Save that cost by using the most widely used server I/O: standard Ethernet.

Bookmark and Share

Top 10 Reasons to visit Xsigo at VMworld

August 10th, 2010

This post was written by Jon Toor

10) Win a trip down coastal Hwy 1 in a Bugatti Veyron, the world’s fastest production car

How about a spin in the world’s fastest production car? We’ll have live drawings daily for three lucky winners who will have the opportunity to travel down coastal highway 1 in a Bugatti Veyron. At 1001 horsepower, there’s nothing else like it. Go home with a VMworld experience you’ll never forget. Enter at booth 1031, next to VMware. Click here to pre-register and triple your chances to win.

9) Manage your private cloud like never before

See a new, more intuitive way to manage server connectivity. Xsigo’s new UI incorporates the concept of resource clouds, so you can connect VMs to functional groups of storage and network resources, such as an Engineering storage pool or DMZ network. It sure beats managing ports. Come see how this can simplify your life.

8 ) Manage server connectivity on an iPad

Nothing could be easier (and more cool) than Xsigo’s new iPad-based management app. We’ll have iPads at the booth (#1031, next to VMware) so you try managing virtual I/O resources simply by touch.

7) See the integration with vCenter

We’ll have lots of servers in the booth all running the latest ESX so you can see the integration with vCenter. View and manage virtual I/O resources on a tab within vCenter.

6) See the world’s fastest server I/O

If the Veyron is the world’s fastest car (it is), then we should have the world’s fastest I/O to go with it. We do. At 40Gbps per server connection, Xsigo’s new QDR InfiniBand server links will let you obliterate I/O bottlenecks.

5) Win a Sony PS3 + Racing Wheel

If you prefer to take a prize home rather than take a spin in the Veyron, you’ll have the option to choose a cool gaming system: a Sony PS3 and Logitech racing wheel and pedals. Live drawings daily. Click here to pre-register and triple your chances to win.

4) Meet with Xsigo engineers and PMs

Technical experts will be in the booth (#1031, next to VMware) at all times, including developers who designed the product. Get direct answers to your toughest questions.

3) Attend the Xsigo/Salesforce.com breakout session: Managing cloud resources

Learn how virtual I/O helps manage resources data center-wide. Matt Cowger of Salesforce.com and Cam Ford, Xsigo’s Director of Product Management will present a technical review with  examples to highlight the key points. Wednesday at 4:30PM.

2) Experience Xsigo in the VMware infrastructure

Learn how Xsgio complements VMware in their own VMworld 2010 infrastructure. Stop by and find out why VMware uses Xsigo to do more with virtualization.

1) See the most exciting development ever in converged I/O

Be a part of history. Experience the most exciting development ever in converged I/O. Visit Xsigo and learn how the world of server connectivity is changed for good.

Bookmark and Share

Why Marc Benioff, Vinod Khosla, and Ray Lane Invested in Xsigo Virtual I/O

August 8th, 2010

This post was written by Ashok Krishnamurthi

Marc Benioff, CEO of Salesforce.com, recently invested in Xsigo. He joins a group of other savvy, high-profile investors including Vinod Khosla of Khosla Ventures, and Ray Lane of Kleiner Perkins, so he’s in pretty good company. Why would these guys — all of whom are adroit at spotting the “next big thing” — finance a data center infrastructure company when cloud computing is emerging as the future of IT?

There are three reasons why I believe Xsigo is a great bet for them.

1) Xsigo is an integral part of the cloud revolution. Inside every cloud you’ll find physical data centers. To reach the full potential of the cloud we are transforming the way those data centers are designed and run. Xsigo delivers a key part of that transformation, the technology needed to make data centers cloud-ready.

2) Xsigo is proven. Our gear is in production use at Fortune 500 and Global 2000 companies such as Salesforce.com, Disney, Accenture, and Capgemini, VMware, EMC, and many others.

3) We are about to assert our leadership even further. At VMworld 2010 we will announce a new technology that every enterprise data center manager will need to consider.

So what makes Xsigo’s technology so compelling?

Building a more flexible data center

The problem Xsigo solves is data center inflexibility. Today’s data center technologies were not designed for the cloud. They were designed for the way we used to run IT, with servers, storage, and networks, interconnected by hard wires and fixed resource mappings.

For the cloud we want a resource-on-demand, utility model — an agile design that allows assets to be configured and then re-configured as needed. Whether the data center is hosting a public cloud such as Salesforce.com or a company’s own private cloud, that agility enables the flexibility and fluid scalability that make the cloud concept so powerful.

Xsigo has the technology to do this, and can do it in a very non-disruptive way. Unlike previous approaches to the agile data center (remember “grid computing”?), Xsigo fits in with existing server, networking and storage technologies. And it fully complements server virtualization.

Why Xsigo virtual I/O

Xsigo’s sells virtual I/O, a technology that brings data center connectivity under software control. That may sound a little abstract, but it’s actually pretty straightforward.

You’ve seen the typical data center with cables running everywhere. An enterprise data center is inherently complex, with servers, network switches, and data storage devices interconnected by miles of cables that are weaved into racks, under floor tiles, and through overhead cable trays. It’s a management challenge to install. And once completed, it’s very costly to change.

That inflexibility was OK in the pre-cloud era. Back then, servers ran just one application for their entire five-year life, so connectivity was simpler and far more static.

The cloud is real. And it changes everything.

In the cloud era, we place new demands on server connectivity. Now a server may run 20 different applications, which necessitates far more connections to networks and storage — 10 to 20 cables per server is common. The resulting complexity is not only expensive, but is also hard to change. Adding new servers takes more time, and reconfiguring existing servers is nearly impossible. All of which impacts scalability and flexibility, two core values of the cloud.

In a cloud infrastructure, the demands on a server will definitely change over time. Data center architects try to anticipate requirements when they design an installation, but doing that accurately over the life of a server is virtually impossible. A company may grow, enter new lines of business, or undergo a merger, all of which will change IT demands. Compliance requirements change. HIPAA, SOX, and payment card industry rules, for example, impact data center design.

And then there is technology change. During the next five years (the life span of a server installed today), there will be at least three new network types introduced. What does the data center manager do with the existing servers? Connect them to the new technology (a costly project), or leave them off the new networks and thereby limit his flexibility? Neither is an attractive option.

The concept of the cloud is to accommodate change.  But to make change truly seamless we need a better infrastructure model – a more flexible data center foundation. And that’s what Xsigo is doing.

Xsigo lets you connect any server to anything in the data center, and then change those connections anytime you want… in software. And Xsigo consolidates those connections to just one cable. It’s the equivalent of having up to 64 cables inside one. Need to change the configuration? It’s a simple software command that can be done at any time without ever touching a wiring bundle or taking a server down.

The next wave in data center design

The implications of this are huge, big enough in fact to attract some serious competition. Cisco and HP both have virtual I/O technologies built into their server product lines. But Xsigo’s answer is better specifically because it’s not built into a server. It is open, which means you can connect all of your servers from all vendors to a single virtual I/O management environment – an infrastructure cloud.

For cloud providers, this can help enable new service offerings, rather than presenting a barrier to them. And for enterprise data centers, it complements server virtualization to create the other half of the private cloud; it is the flexible, virtualized infrastructure that meshes with the virtualized servers.

All of which is why I believe this is a great investment.

Investors look for the white spaces on the technology horizon, where new products or services can create a richer picture or fill in the gaps. Cloud computing has created a richer picture, giving companies new options for IT. But it’s also created a gap: the need for an infrastructure that’s as flexible as the IT model it supports.

In Xsigo, these investors saw the opportunity to be a part of the cloud revolution, filling a gap that the cloud model created. And in the process, a chance to be a part of the next big thing in the data center.

 
 
Bookmark and Share

InfiniBand: Faster AND Greener??

July 24th, 2010

This post was written by Jon Toor

Prius racing?? I’ve heard of it being done. I’m told that gas mileage gets factored into the race standings, but it sounds pretty dull. The fact is that “green” and “speed” don’t usually go together.

Which is why I was intrigued by two InfiniBand articles that recently appeared. Taken together, they report that IB is both the fastest server interconnect, and has the greatest potential to increase data center power efficiency as well.

Here’s what they said:

1)   The Register reported in this June 29 article, titled InfiniBand to outpace Ethernet’s unstoppable force, that “InfiniBand, more than Ethernet, has kept pace with the exploding core counts in servers and massive storage arrays to feed them, which demand massive amounts of I/O bandwidth in the switches that link them.”

Not only has IB been ahead on past speed transitions but already has specs in place for speeds to 312Gbps, and for  new encoding mechanisms that will use that performance much more efficiently, consuming just 3% of the bandwidth for overhead.

OK, we knew IB is fast. That’s why it is used in 37% MORE of the TOP500 supercomputing sites than it was just one year ago. But green, too?

2) Google researchers reported an energy savings analysis in a  June 19 journal paper, titled Energy Proportional Datacenter Networks.

When you’re building data centers near dams, as Google is, you’re probably doing some serious thinking about saving power. So what green technology did they reveal in this paper? InfiniBand.

They discussed the concept of “energy proportional,” which essentially means, “use less power when you can.” Their finding was that servers are getting better in this respect, but that networks are not. In terms of power, networks “are always on regardless of whether they are flowing data packets, because they must still send idle packets,” they report.

The implication of this is that “the datacenter network can become a significant fraction (up to 50%) of cluster power.”

The answer is to design a network whose power consumption is more proportional to the amount of traffic it is moving.

Here they found that IB technology has designed-in capabilities ideally suited to this power-saving goal. The IB architecture includes provisions that make it possible to shut down inactive lanes — and thereby reduce power — when they’re not required. Power savings in a scaled-down performance mode can reach 60%, enough to save $3M over four years in a 3400 node data center.

Faster and Greener

So it is possible to go fast and save energy doing it. Which I guess isn’t too surprising. Porsche after all is showing a hybrid that does 0-60 in about 3 sec.

But that car will cost over $600K. That same money will buy about 1,000 of the world’s fastest server I/O cards.  And they just happen to be the most efficient ones, too.

InfiniBand to outpace Ethernet’s unstoppable force

 
 
Bookmark and Share

Getting over the cloud… the preconceptions, that is

July 9th, 2010

This post was written by Jon Toor

Can a cloud model surprise you… in a good way?

We all hate surprises. Too often they’re bad things. But occasionally you can get surprised in a good way. That’s what happened to an IT manager I ran across this week. He achieved some remarkable results in a way he did not expect: in the cloud.

His company, a large financial services provider, needed to improve the scalability of an online service offering. Batch jobs were already running too slowly with the number of users they had. And they were expecting the user base to grow significantly. What would they do then?

A cloud service provider thought he could help, but the IT manager was skeptical. Would it be fast, secure, reliable, etc?

The cloud provider was able to satisfy the security and reliability concerns by explaining the architecture. The underpinnings relied on Xsigo virtual I/O, so they were able to provide isolated connectivity to servers. It also let them configure redundancy and fast failover throughout the infrastructure — whether a connection or an entire server failed. It helped that the provider was an established managed hosting company, so they were a known quantity from a business perspective.

But that still left the performance question. And the only way to answer that was to try it. You can talk about car engines all you want, but until you take the car to the track, you won’t really know its top speed.

So the cloud provider set up a test: a batch job with data from 100,000 users. This job that took 2 1/2 hours to run in the IT manager’s own shop (their forecasted number of users was 400,000, so you can imagine why the IT manager was anxious for a new solution). The goal was to drop that to 1 hour. That would be a success and would allow the IT manager to safely expand his user base.

When they ran the test, the surprising result was this: it took six minutes. 96% less than the original, and 83% less than the goal.

That’s good, but would it scale? They tried it again with a much larger data set and found that it scaled nearly linearly.

It’s not often we’re surprised in a really good way. And with all the fear, uncertainty, and doubt around the cloud (the pundits see all kinds of threats), it’s pretty easy to believe that any surprises are likely to be bad ones.

But if this IT manager had gone with that flow rather than looking at the potential upsides, he would have missed a solution that saved his company significant time and money.

He had a healthy skepticism… which is a form of preconception. But he got over it.

 
Tags:
 
Bookmark and Share

Does the World Need Virtual I/O?

July 2nd, 2010

This post was written by Jon Toor

Network Computing.com recently raised this question: “Does the world need virtual I/O?” Why introduce virtual I/O, they ask, when 40G Ethernet will provide plenty of bandwidth for any server?  Isn’t virtual I/O really needed only for bleeding-edge, large scale applications?

Rather than respond here myself, I put the question to two data center architects who use virtual I/O. Neither runs bleeding-edge apps — they provide web services, database, Exchange, all the usual stuff. And neither operates an esoteric Star Wars-type computing environment.  Both run typical data centers in facilities familiar to us all.

So why do they need virtual I/O? Here are their answers.

Case 1: Distribution Company

This company is a distributor in the Midwest. The IT department maintains databases for orders, inventory and customer relationships. And they provides similar IT services to their clients to help them streamline their own operations.

So what do they get from virtual I/O? Here are a few key points:

Performance: Increased bandwidth was an important feature for them. Like a lot of shops, they’ve consolidated, so they needed to get more out of fewer servers. They’re running high-end Nehalems and very fast storage, so the opportunity to get 40Gb through a single server link was important to maximize sever utilization. And they wanted to do it today, not sometime in the future.  The fact that Xsigo’s IB link delivers by far the industry’s lowest latency (in the nanoseconds, as opposed to microseconds or even milliseconds for FCoE), helped too.

Agility: They are building what they describe as a private cloud. They want to run any application on any server, so the ability to dynamically manage connectivity is important.

Cannot rely on VLANS: They absolutely cannot run a single Ethernet connection into a server and then carve it up. Multiple physically isolated networks are essential — VLANS alone do not provide the isolation they require. Within their cloud there will be multiple businesses (both internal and external) sharing the resources. There will also be multiple data types, including credit card information. From a legal and a compliance standpoint, network isolation is required.

Future Proof: Virtual I/O makes change simple, whatever that change might be. As the company’s architect pointed out, “The future is hard to predict. What happens if I suddenly need to connect my legacy SAN to the cloud? Or if I need a new network, or FCoE enters my plans? With virtual I/O that’s all easy to accommodate.”

Case 2: Online Transaction Company

The second architect, at an online transaction firm, had a slightly different spin on virtual I/O and how it helps. His company is larger and more complex, so he values  tools that help reduce complexity.

Simplified planning: In a complex environment, just planning cabling can be a challenge. With virtual I/O they don’t have to think about it because every server is cabled the same way. The architect commented, “I don’t have to think about whether it’s fibre, single or multi-mode, Cat 6, or whatever. It’s all the same.”

$1.5M cost savings: These guys know data center design, and know very well how to save a buck. But their next-best alternative to virtual I/O would have cost $1.5 million more on day one.

Agility: The ability to change things rapidly with virtual I/O can help avoid downtime. As the architect pointed out, “It’s not that we’re changing things frequently. But when we need to change something right now, it needs to be done — right now!”

Remote management: Because this firm operates multiple data centers, they need remote management. “If I can fix an issue without having to get a guy physically to the data center, that’s a huge time savings.”

Cannot rely on VLANs: Like the distributor, this company has compliance regulations and payment card industry rules to worry about. Not to mention their own peace of mind. The architect pointed out, “I manage my shop tightly to ensure security. You won’t be seeing my name in the news.”  Isolated networks are employed wherever needed, and virtual I/O lets them maintain that isolation while still getting the flexibility to run any app on any server.

“To us, virtual I/O improved the quality of life,” the architect concluded. “We simplified planning, accelerated critical management tasks, and saved cost at the same time.”

That’s live input, this week, from two IT managers on the front lines of data center operations. Neither is unusual in any way except that they looked for and found a better way to manage their infrastructure.

Do they need virtual I/O?

So, getting back to the original question, do they need virtual I/O? No, of course not. They both ran their data centers just fine before they installed Xsigo.

But does having virtual I/O give them a competitive business advantage? And does it improve their quality of life? Absolutely.

Bookmark and Share

The Explosion at the Other End of the Cloud

June 18th, 2010

This post was written by Jon Toor

We talk about cloud as a data center concept, but what about at the other end of the pipe?

Just as the cloud is revolutionizing the data center, there is an equally dramatic revolution going on at the the content consuming side of the compute world.

Namely, the trend towards mobile devices is emerging as a force.

An article in last week’s Newsweek magazine offered some interesting market numbers about Apple’s revenue split for desktop verses mobile products. PCs no longer dominate at Apple. It’s not even close. The company’s PC business has dropped from 70% of  revenue to less than 30%. (During that time, Apple’s share of the overall PC business has remained roughly flat, at about 8%.) In the PC’s place, mobile now rules the Apple roost. The iPad is selling a at a rate of one unit every 3 seconds. iPhone 4 pre-orders recently  brought ATT’s system to its knees.

While mobile is not yet the dominant model for the compute world — tablets have a tiny share of the overall market — it’s a trend that has a lot of interest.

The same Silicon Valley venture capitalists who invested early in the internet now are doing the same with tablets. John Doerr of Kleiner Perkins (a Xsigo investor) commented about iPad, “It is not a big iPod. It is a very big deal.” And he backed that up by doubling his firm’s iPad-related investment fund from $100M to $200M.

For data center managers, the implications are huge. Growth at the consuming end of the cloud can mean only one thing: equivalent growth at the distribution end of the cloud. Data center managers will be in the drivers seat as consumers seek better and richer ways to use these devices.

The trends towards mobile devices and VDI point to increased demand on data center managers to be responsive. More importantly, it’s a great opportunity to make the shift towards an inherently responsive infrastructure to make that job easier and far more cost effective.

 
 
Bookmark and Share

Cloud Computing Shopping List: 4 Key Ingredients

June 11th, 2010

This post was written by Jon Toor

Defining the cloud remains a challenge. But we may be getting closer to agreement on the cloud building blocks.

CIO magazine this week provided a list of four ingredients that make up the cloud. Virtual I/O was part of that list. Here’s a quick look at their Fab Four:

** For more on this topic, join our Xsigo/Dell webcast, “How Virtual I/O Lets You Build a Private Cloud with the Vendors of Your Choice,” June 17, 1PM Eastern. Click here to register. **

1. Application Integration

In this context, “integration” is intended to work towards IT as a service, as opposed to simply managing hardware that hosts applications. “Integration” could mean anything, but the article provides a specific example: Enhancements such as administering user identities across multiple applications from a single location can greatly streamline management.

2. Security

A critical element, but one that’s still a work in progress, according to this article. Which partly explains why we see people working towards a private cloud as a first step.

3. Virtual I/O

There are multiple reasons why virtual I/O is essential to the cloud: performance, real-time management, simplicity, etc. But for the IT manager (and Xsigo user) quoted in this article, it was all about scalability.

“When you’re in the dev/test stage, having eight or 10 [Gigabit Ethernet] cables per box is an incredible labeling issue; beyond that, forget it,” says the IT manager. “Moving to virtual I/O is a concept shift—you can’t touch most of the connections anymore—but you’re moving stuff across a high-bandwidth backplane and you can reconfigure the SAN connections or the LANs without having to change cables.”

4. Storage

Storage is an issue because virtual machines require physical storage, just as physical servers do. While the article highlights this as a cost driver, remember that it’s an I/O driver as well. The fact that virtualization is storage intensive implies that it’s I/O intensive as well. Remember, with VMs most of that storage will be external to the server.

While there may not be a single generally-accepted definition of what the cloud looks like once you’ve arrived, there is a lot of value in the conversation about how you get there.

 
 
Bookmark and Share

Army Attacks the Cloud!

June 3rd, 2010

This post was written by Jon Toor

Still not convinced the cloud is real? Then consider three cloud announcements in the news this week. One from the US Army, one from Microsoft, and the last from HP.

In this Information Week article, the CIO of the Army discusses their plans to consolidate servers by creating “compute clouds in select data centers.” The Army is one of the world’s largest IT operations, with a $10 billion IT budget and 1.4 million users.

The Army’s plan of attack reads like Cloud 101.  It wants to “exert control over server deployments as it prepares to consolidate data centers and, in the process, convert designated data centers into cloud computing environments that provide shared services across its operations.”

Microsoft also made news, announcing a joint cloud computing center with Taiwan’s economics ministry on Thursday at the Computex electronics show. And they announced plans to work with two local companies on new designs for servers meant specifically for cloud computing.

And finally there was the HP announcement of their plan to restructuring IT operations around the cloud.

All of these point to the immediacy of the cloud transformation. Why? Because the cost savings potential is just too compelling to ignore.

We work every week with IT shops saving hundreds of thousands of dollars in immediate capex by adopting cloud models with virtualized servers and virtual I/O. Most compound those savings over time with increased operating efficiency. A public cloud provider we worked with recently saved over $1,000,000 in immediate capex by moving to virtualized I/O.

So the choice will become clear. You can streamline your own operations with virtualized servers and I/O. Or you can outsource to a public cloud provider (who is probably virtualizing their servers and I/O). Or you can ultimately face the question of why your IT operations are not cost competitive. The Army saw this challenge and opted to attack the cloud hill.

Bookmark and Share