CloudBzz

Skytap is known as a cloud dev/test provider today, but they have been seeing more workloads coming on-board including ERP migration, training, demos, etc.  So perhaps they are not as targeted as we think.  This can be a risk, where customers start to wonder what you stand for.  Skytap entered in the dev/test market, and they did not have a grand plan to expand from this.  It’s being driven by customers. So,

Today they are announcing “multi-network” capabilities to enable multi-level network topologies to be as flexible as your on-premise networks.  Only it’s a lot easier – you can deploy this in a browser.  It even allows you to save configurations and check them in and out of the Skytop repository.  This is their “virtual private cloud” capability.  This goes significantly beyond the current Amazon VPC solution with much more flexibility and configurability.  It’s a lot less work to set this up than with Amazon VPC, which basically assumes that a developer is part of the network team. 

Skytap is basically claiming the ability to enable the kinds of networks shown above.  This is a nice differentiator and makes it easier for enterprises to move multiple complex workloads, like SAP and other multi-tier applications, to a cloud.

I think we might be at the very beginning of an interesting new phase in the evolution of cloud computing — regional and local clouds.  Local and regional hosting is nothing new – there have been smaller players operating in the shadows of the big hosting companies for years.  Some of these organizations are resellers of larger data center capacity, while others have their own facilities.  It’s only natural that some of these local hosters may start new cloud initiatives to keep their customers from ending up at Amazon, Rackspace or Unisys.  Some will be successful, while others will fail, but no matter how it turns out – there will be a robust local cloud economy.

I spent some time with Jason Baker and Johnny  Hatch from ReliaCloud last week while I was out in Minnesota.  ReliaCloud is an offshoot of Visi, a St. Paul-based hosting and colocation provider.  Recognizing that the traditional managed hosting and colocation market is being eclipsed by cloud computing, the Visi team decided join the cloud party.  They decided to offer their cloud under a new brand, ReliaCloud, as a way to be more interesting to the market.  Here’s what else I learned:

• The value proposition for a local cloud is strong.  Customers who want to keep tabs on where their applications are running, need to be able to audit their cloud service provider, need custom configurations, or just want the extra comfort of knowing the people who are running the cloud, will find a local cloud to be attractive. In some cases, regulatory issues favor a local cloud as well.  In healthcare or financial services, the fact that you know for sure where your data is at all times is comforting. 
• Building a local cloud is not a huge challenge given the tools available today.  ReliaCloud uses VMOps, a venture-backed cloud stack provider out of Cupertino, CA.  VMOps has a pretty strong solution that allowed ReliaCloud to get up quickly with a minimum of hassle.  Jason and Johnny both said that VMOps was very responsive and delivered on all of their commitments.  Other tools that ReliaCloud evaluated included VMware, Xen Cloud from Citrix, Eucalyptus and others.  They just felt that VMOps was stronger and offered more of what ReliaCloud was looking for.  They really like how VMOps deals with networking and storage, and how robust their HA VMs are (based on Xen this release but with plans for other hypervisors in v2).  ReliaCloud built their own front end for self-service on top of VMOps APIs, though they commented that using what VMOps had out of the box would have saved them a lot of time.  They use the Tucows Platypus billing system, monitor their cloud with Nagios, and their storage is based on OpenSolaris ZFS managing Dell storage shelves (VMOps is adding more enterprise storage options too). 
• A local cloud can have quick success, especially if you already have customers you can turn to.  ReliaCloud has been in beta for a little over 2 months.  During their beta period, ReliaCloud is free and nearly 100 customers have signed up.  In some cases, customers are running live production systems already and getting a lot of the benefits of cloud computing early on.  ReliaCloud is not sitting still either.  They just hosted a joint seminar with enStratus (Minneapolis-based cloud tool provider), are offering free cloud computing to nonprofits (doing good for good PR), and are exploring providing private label cloud services to integrators and other hosters who don’t want to do it themselves.

ReliaCloud is still early in their journey to the cloud, but are already having great success (I happen to know that in December one of the major vCloud-based telco clouds in the U.S. had less than 10 customers using their service).  An analogy I heard recently comes to mind…   Think of the big hosting companies as giant boulders.  They take up a lot of space, but they leave a lot of space between them for rocks, stones, pebbles and sand.  Local and regional clouds are there to fill the empty space, and there are a lot of them.

2009 has certainly been a cloudy year.  The sheer volume of real innovation somehow makes all of the hype worthwhile.

While there were many companies doing interesting and innovative things in the cloud – Microsoft Windows Azure could be a strong 2010 contender – the decision on who wins for 2009 is no contest.

Amazon gets the CloudBzz “Innovator of the Year” award with a never-ending stream of great stuff that only seemed to accelerate as the year progressed. Here are just some of the great announcements out of the big river in 2009:

January- Management Console makes it easier to manage your instances.

February – IBM Adds DB2/WebSphere AMIs and starts endorsing AWS for the enterprise.

March - Reserved Instances results in lower pricing, Eclipse support is added to make it easier for larger Java projects.

April - Hadoop support is added as Elastic MapReduce which builds on one of the most strategically powerful ways enterprises can use AWS, IBM support is productized with a range of offerings.

May - In a big move that impacted RightScale, Amazon added a host of needed features including auto-scaling, load balancing, and real-time instance monitoring with CloudWatch.  Recognizing the limitations and costs of moving terabytes into AWS over the Internet, physical data transfer is supported for customers with large data sets.

June & July were pretty slow – minor announcements only.

August - beyond dropping the prices for Reserved Instances, the Virtual Private Cloud announcement shakes up the enterprise cloud market.  They also make some nice moves in security with rotating credentials and multi-factor authentication.

September - mostly upgrades to earlier releases, though their Solution Providers Program is a nice boost for partners.

October - Amazon announced lowered the pricing for EC2 instances (price war heats up), added instances with high (though not really high) memory, and my favorite announcement of the year (closely beating VPC) — Relational Database Service.

November - Amazon got their SAS70 Type II audit done (big whoop), expanded into Asia, and added a .NET SDK to counter Microsoft Windows Azure.

December - pricing changes, EBS-based EC2 boot, and  a few other announcements were nice, but EC2 SPOT INSTANCES really shakes things up. In addition, Virtual Private Cloud moves to unlimited beta (previously limited beta).

I had a “discussion” on twitter a few weeks ago where I predicted that Microsoft’s Windows Azure would be “the one to beat” in the enterprise. It’s nice that companies are using Amazon and other clouds, but for the 80-90% of Windows/.NET applications that run your typical enterprise, Azure will be king.

I’m at the PDC in LA today and in packed sessions of enterprise developers and ISVs who are genuinely excited about moving their Windows workloads to the cloud. Azure is not targeted towards the big SaaS/Web 2.0/Facebook app crowd. Instead, they are going after the enterprise users who drive the bulk of spending in the tech market.

Consider the following factors:

1. Microsoft Owns the Platform: Sure, most enterprises have used VMware to virtualize their Windows apps, but they’re first and foremost Windows apps.  VMware should be afraid of losing this dominance as workloads move to the cloud.  After all, if you’re not managing the virtualization, do you care who provides it (as long as it works and is secure)?
2. Microsoft Owns the Tools:  Nearly all Windows apps are written with Visual Studio and most run on SQL Server.  You can guarantee that there will be tight integration between the Azure platform and the Windows tools.  Deploying your apps to Azure will be WAY EASIER than deploying to any other cloudd.
3. Microsoft’s ISV Base is Huge: Did  you know that most SAP installations run on Windows?  If  you sell business apps to mid and large enterprises, you probably run on Windows first.  Microsoft is aggressively pursuing ISVs to move to a SaaS model on Azure with great tools and support that will only get better.

Face it – despite their very slow start, Microsoft knows that Azure is a make-or-break situation.  They are innovating, listening to customers, and pouring $ billions into the cloud. They will be happy to let the Web cloud crowd use Amazon, Google and others.

What does this mean for VMware and the VCE Consortium?  I actually believe that VMware will be a survivor in this business, but perhaps with less influence than they have today.  There will always be a need for higher-end solutions for complex deployment scenarios, and VCE is well-positioned for this.  Where you need a lot of customization, transition services, and other help for your more complicated applications (like the aforementioned SAP), a VMware-based cloud will be a great fit.  The topology for a typical SAP application, for example, may be challenging for Azure but is perfect for a VMware cloud (such as Verizon, Unisys, Terremark, and others).  And VMware has a lot of friends.  Lastly, VMware can make workload migration from like-to-like hypervisors easier than moving from them to Azure.  Don’t count them out!

Azure’s not perfect (SQL Azure has serious scaling challenges for analytic workloads), but it’s a strong solution for the bulk of those 2,000-4,000 regular applications that run a business.  And that’s just fine with Microsoft.

Today RightScale did a webinar on their Cloud Business Intelligence offering with Talend, Jaspersoft and Vertica.  One of the bigger objections to cloud BI in the past has been security — how can I move all of this mission critical data to a public insecure cloud?

With Amazon VPC now in the picture, the BI datasets are now as secure at Amazon as they are in your data center.  Why wouldn’t you use the cloud for your BI needs?

Last night Amazon announced the most significant cloud development of 2009 – the Amazon Virtual Private Cloud (VPC). The AWS Developer Blog version is here.  The importance of VPC cannot be overstated.  It will literally change how enterprises think about public cloud providers and the opportunity to gain efficiency and flexibility in datacenter operations.

By integrating with the security, governance and compliance infrastructures of enterprise IT, VPC eliminates one of the primary barriers to cloud adoption for mainstream business computing. Sure, there are still going to be issues, but this was the big one.

I won’t rehash all of the offering details here.  You can read them on Werner Vogels blog and TechCrunch.

The hybrid cloud is a reality.  You can now integrate your internal fixed IT infrastructure with large external clouds with a high degree of integration with enterprise tools. VPC allows you to assign IP addresses, create subnets, and connect your existing data centers to Amazon using secure VPN technology. Sure, this is not the same level of connectivity via dedicated secure lines that most big outsourcers provide, but it’s pretty strong and many very smart people (including Chris Hoff at Cisco) are bullish from a security perspective.

I will think about this a bit more, but Werner Vogels makes the claim that “private clouds are not clouds” mainly because they are not truly elastic.  There may be some benefits to using Eucalyptus or VMware’s vSphere in your data center, but you still need to buy hardware and install it and that’s not cloud computing according to Vogels.

One thing that’s certain, the game has changed – again!  Amazon’s VPC is far and away the most significant cloud computing announcement so far this year, and I’m going to go out on a limb and predict that on December 31 it will still hold that distinction.

What do you think??

Update: InfiBase has ceased operations, but the analyses they are providing may continue.  Stay tuned.

A stealth start-up called InfiBase has published some very interesting data on their blog recently. It makes me want to know more about them, so if you have the scoop let me know.

First, they have put out two posts on sites using Amazon EC2, with other cloud providers included in the last posting earlier this month. Here is their chart showing the top 500,000 sites by cloud providers.  Note how close Amazon EC2 and Rackspace CloudServers (based on Slicehost) are in this ranking.

Source: InfiBase

I was interested to see Joyent in third place, well ahead of both Google and GoGrid, and I wonder what this might look like a year from now.

In another post InfiBase performed a deep dive into the processing dynamic of various EC2 instances, including which processors are being used and how they stack up.  Here is just one of their great charts which shows that AMD processors are used at the low end of EC2 while Intel takes over at the very high end.

Source: InfiBase

With the data they are previewing in their blog (see the full posts there), I am intrigued.

As a general rule, I am happy to count Salesforce.com as a cloud computing company.  They really made the SaaS market what it is today, and their Force.com platform-as-a-service was a great innovation.  They are not an infrastructure cloud provider like Amazon, Rackspace or others, but okay – they’re a cloud company.

However, when I see their current marketing and branding it makes me want to chuckle.  Instead of Salesforce, Successforce, and Force.com, they now market Sales Cloud, Service Cloud, and Custom Cloud.  They already had the cloud creds, but by trying so hard it makes them look a bit silly.  I wonder if this rebranding is hurting or helping their sales numbers…

Following Unisys’ announcement on Wednesday regarding their cloud computing strategy (Unisys Moves to Break Through Barriers to Adoption of Cloud Computing) I had the opportunity to speak with Rich Marcello, president of Unisys Systems & Technology, and Sam Gross, VP of Unisys Global IT Outsourcing Solutions. What struck me was the coherence and clarity of their cloud computing vision as compared to HP or IBM.

Unisys’ strategy bridges public, private and hybrid cloud models, and includes well-differentiated infrastructure, platform and software as a service offerings (IaaS, PaaS, and SaaS). Further, they wrap this all in a set of comprehensive service offerings that they can deliver globally. It’s a big vision, and if they can pull it off it should make them one of the more interesting providers out there. One of their key differentiators is Unisys Stealth (described below).

The Big Picture

As you see here, Unisys is providing a pretty comprehensive vision.

Click To Enlarge

1. Private Cloud – Unisys will turn your existing data center assets into a private internal cloud with all of the things you’d expect – virtualization, automated provisioning, etc.
2. External Cloud – through a network of five data centers owned an operated by Unisys, they will provide companies with external cloud computing environments. This is not really a public cloud in that people can’t just come in and sign up. It’s part of their managed services portfolio and is tightly aligned with their private cloud solutions.
3. Cloud-in-a-Box – due later this year, Cloud-in-a-Box will be a pre-configured cloud infrastructure delivered either as software or with a software+hardware model. This is like Joyent or Enomaly.
4. Hybrid – early next year Unisys plans to provide a hybrid cloud model where your applications can run on both internal and external cloud resources. This is a virtual data center model.

XaaS – X as a Service

To my knowledge, Unisys is the first to publicly support having a complete set of offerings for XaaS. Amazon is really IaaS, and Google is more of a PaaS play. Salesforce is a combined PaaS/SaaS offering (PaaS for force.com and SaaS for their SFA/CRM applications). Even within these offerings, Unisys has some interesting differentiation.

Click To Enlarge

1. IaaS – most public clouds are built on a “scale out” model with many smaller machines. In addition to “scale out,” Unisys also includes “scale up” capabilities to allow you to use large, enterprise-class hardware in your cloud (for running databases, for example). What’s important about this is that Unisys allows you to provision servers that match the software licenses you already own. If you have an Oracle database license for an 8-way Solaris server with 8GB of RAM, you can’t use that on Amazon’s EC2.
2. PaaS: Java and .NET – Unlike Salesforce which requires you to program to their model, Unisys provides a container model to allow your Java and .NET applications to run in the cloud. You write the application and deploy into these containers – you don’t have to install or manage the core components (e.g. WebSphere and DB2 or Windows and Oracle), you just deploy your applications. This should appeal to IT.
3. SaaS – while the offering is a bit limited at this point, Unisys is offering a set of end user business applications in their cloud (email, SharePoint, virtual desktops).
4. My Secure Application as a Service: AaaS – this is where an existing application you have deployed can be moved into the cloud and managed from a provisioning and security perspective.

People Services

Unisys claims to have an 800-person services organization that can provide a complete range of assessment, advisory and implementation services to their customers. This is in contrast to what they termed “do-it-yourself” clouds from the other guys.

Click To Enlarge

Unisys Stealth – A Secret Weapon

As has been reported in many quarters, security is the #1 deterrent to more enterprise cloud adoption. While Unisys uses standard enterprise-grade security technologies you’d expect to find in a large IT shop, they also have a unique solution called Unisys Stealth (link goes to a paper describing Stealth that is specifically targeting the Defense industry). Announced last November, Stealth is a network appliance that makes data and even hardware (desktops and servers) invisible to network sniffers and other similar technologies.

Stealth works at the link layer (layer 2) of the TCP/IP stack, which means that every packet on the network is cloaked unless you have the right key.

“The result is a cloaked network that secures data-in-motion
and hides servers and PCs in plain sight. Devices that do
not have the same workgroup key remain cloaked from
unauthorized eyes. Without the correct key, users cannot ask
for the data from the server or send data to the server or
workstation. They can’t even ping the server or workstation.”

Click To Enlarge

So now you can run any application in the cloud and only authorized users (controlled by your network administrators) can access the information. Stealth even applies to your storage infrastructure — your SAN becomes invisible and therefore inaccessible to hackers. All of this is accomplished without any changes to your application. You add the appliance to your network, Unisys deploys it in the cloud, and voila – instant cloaking.

Conclusion

Unisys has articulated a comprehensive and cohesive cloud computing vision, while simultaneously addressing security in a new and very powerful way. If they deliver all that they are claiming, Unisys should be well-positioned for success in the great cloud migration.

I got a chance to spend time with James Duncan and Bryan Bogensberger of Joyent at #e2conf today.  I’ve always been a little bit cloudy on what Joyent actually does.  Their Web site (which is getting a makeover) is not all that clear today.

Let’s take a look at their stack as they described it to me this afternoon.

Accelerators – these are roughly equivalent to Amazon’s EC2 instances, though with a lot of compelling differences (if you’re using/considering EC2, you should check out Joyent).

“Cloud Control” – this is not a product per se, but it’s their management interface into their Accelerators.

Smart - this is their Javascript-based platform-as-a-service (PaaS) offering from Joyent’s acquisition of Reasonably Smart (where James and Bryan hail from). The value of Smart to Joyent is as a feature of their stack, and not as a compelling standalone solution. It’s going to be open-sourced so you can run the Smart PaaS offering inside your firewall. I’m a bit dubious on whether or not they will see much adoption, but perhaps it will kick butt.

The Hybrid Cloud

What struck me as totally unique and compelling about Joyent was an aside comment that they have a couple clients today running the Joyent Cloud Control management interface across both internal (IT-controlled) and external (Joyent-controlled) infrastructure. The net effect is that, to the IT user, their own servers and Joyent’s accelerators combine for a seamless and burstable hybrid cloud.

Effectively, your servers are “joined” to the cloud. This is my “marketecture” view from my conversation with James and Bryan, and what they end up releasing may look very different. But if what they say is true, they may be one of the first to have actually deployed a hybrid cloud into production. That’s huge – like Santa Claus is Real kind of huge!