Hybrid private-public cloud models are the reality in most enterprises in the forseeable future. Developers and business units coninuously go "rogue" and use public cloud services, while IT struggles with maintaining compliance and control and managing ...
Hybrid private-public cloud models are the reality in most enterprises in the forseeable future. Developers and business units coninuously go "rogue" and use public cloud services, while IT struggles with maintaining compliance and control and managing legacy apps in the traditional data center.
I've spoken many times about this constant push and pull between flexibility and control in the cloud. And it is becoming apparent that we need a better way.
Well, today, my friends at Ravello Systems announced that they have launched in public beta their Cloud Application Hypervisor and that they have received $26 million in funding from Sequoia, Norwest Venture Partners and Bessemer.
I've had the pleasure of working with the Ravello team in preperation for this launch, and I believe they have a much needed solution -- built on a very unique technology -- for many of the biggest problems associated with cloud deployment.
What is a Cloud Application Hypervisor?
The best way to think about Ravello's technology is using the familiar hypervisor as an analogy. But while the traditional hypervisor holds a single virtual machine in it, Ravello's CAH holds a complex multi-VM app in it. This allows you to encapsulate a complete application (load balancers, app servers, web servers, databases, etc.) AND it's environment (networking, storage, etc.). The result is complete portability across clouds and between on-premise and public clouds. For example, you could take an existing VMWare-based application running on your data center and deploy it on AWS, Rackspace Cloud or HP Cloud as-is.
Development and Testing in the Cloud
So what is this good for? One of the first use cases Ravello Systems is targeting is the need to do development and testing in the cloud, while running the app in production in the on-premise enterprise data center.
The cloud -- with its unlimited resources and ability to spin up machines quickly and then dispose of them -- is ideal for testing and development. But as mentioned earlier, enterprise IT departments still have many issues with running production apps in the cloud. These issues include compliance, security, cost and fear of vendor lock-in and dependence. With Ravello, developers can deploy the application "capsule" in a public cloud, run multiple instances of it for parallel testing, collaborate on development and generally enjoy the flexibility of the public cloud.
When it's time to move the app into production, IT can simply deploy the encapsulated app in the data center.
In the future, Ravello will address additional use cases such as more general cloud portability, cloudbursting and other scenarios.
How It Works
It's important to note that Ravello is delivered as a cloud service itself. You create an account, log in and can then create blueprints of applications (or use pre-existing ones) which can be cloned and shared.
The leadership team at Ravello brings a lot of credibility to the table. Among them, Benny Schnaider, Rami Tamir and Navin Thadani, were the team that created the standard Linux hypervisor, KVM. They sold the company they created to commercialize it, Qumranet, in 2008 to Red Hat.
Check it out and let me know what you think in the comments.
Zenoss just published the results of an open source cloud survey they did. They polled more than 100,000 of their community members to determine the prevailing sentiments concerning open source cloud deployments, the perceived advantages and disadvantages of this technology as well as to gain insight into future open cloud deployments within IT departments.
Survey respondents included more than 600 IT professionals including system administrators and architects, developers, network engineers and CIOs.
Unsurprisingly, OpenStack appears to dominate adoption plans today but CloudStack and Eucalyptus are on the rise.
Check out the infographic they created from the data below (click to enlarge). If you want the full report from the survey go here.
This coming Thursday, May 10, I'll be giving one of the keynote speeches at the Citrix Synergy 2012 conference in San Francisco. My talk is in the morning and comes right after a distinguished speaker: Sameer Dholakia, GM of the Cloud Platforms Group at Citrix.
You can see the description of the two talks (and the one by Citrix CEO Mark Templeton who speaks on Wednesday) on this Featured Speakers page.
The title of my talk is "From the Bottom Up: Patterns of Cloud Adoption". Here's the abstract:
The current pattern of cloud adoption in the enterprise may surprise you. Rather than big, strategic, top-down decisions set by the CIO, cloud computing services – IaaS, PaaS, SaaS – are being adopted primarily through a pattern of bottom-up adoption. Rank-and-file developers, IT administrators and business decision-makers are embracing cloud services and using them as a way to get their jobs done and drive the outcomes expected of them. In this talk, Geva Perry will explore this phenomenon, including its causes and the implications for the enterprise, as well as for vendors.
Citrix is making a big announcement today. It has two parts. First, it's moving its CloudStack framework, which was owned by Citrix and distributed at CloudStack.org through a GPL license, to the Apache Foundation. Second, it is aligning CloudStack with Amazon Web Services' architecture and APIs.
This is big news that's sure to ruffle some feathers in the cloud computing space. So the questions I have are:
Why are they doing this?
What are the implications?
Why Is Citrix Moving CloudStack to the Apache Foundation?
The move appears straightforward. Citrix acquired CloudStack through its $200 million acquisition of Cloud.com in July 2011. From what I'm hearing, it has had good success with enterprise and provider adoption of the product, but it was far from being accepted as an industry de facto standard, which is what, I assume, they had hoped for. So it would make sense for them to move it to a respected open source foundation like Apache.
But wait. There was already a OpenStack vying for that de facto standard open source platform status, and Citrix announced its support for it two months before the Cloud.com acquisition. The company said it would continue to support both platforms after the acquisition, so what happened?
Citrix claims that the OpenStack foundation wasn't run well: it was dominated by Rackspace and had a "pay-to-play" model. The result was that the APIs were poorly designed and the product lacked stability and maturity (a claim I have heard from others). At the same time, while OpenStack is getting support from the likes of Cisco and HP, the community is somewhat fragmenting, with multiple distributions and extensions from the likes of startups Cloudscaling and Piston Cloud. This started creating a problem for enterprise customers and providers who were getting confused and having a bad experience with OpenStack.
In the meantime, Citrix is feeling competitive pressure from the company it views as its primary rival for actual customers (as opposed to winning the hearts and minds of the "community"): VMWare.
It simply couldn't wait and had to go on the attack with a bold move. This is it -- and it's a pretty good one.
Why Align CloudStack with the AWS APIs?
The final piece in making CloudStack the de facto standard cloud platform is the API. By aligning with the Amazon APIs, CloudStack gains a hghly-adopted, proven API with a massive ecosystem of integrated tools and services around it. They tell me they will have 100% compatibility by the end of this year.
With this move complete, Citrix has a production-proven stable product, which is now an open source platform managed by the widely-respected Apache Foundation, using a popular API with a massive ecosystem around it. Pretty clever, I think.
On the other hand, CloudStack is fighting on three fronts now: its traditional arch-enemy VMWare, the OpenStack camp with the dozens of vendors -- large and small -- behind it, and Eucalyptus. A regular game of thrones. And to paraphrase Game of Thrones: In the Game of Clouds, you either win or you die.
A quick follow up on my previous post on cloud computing adoption patterns. I have been guest blogging on Compuware's CloudSeluth blog and have written a post on this topic titled Cloud and Bottom-Up Adoption. In it I reiterate some of the points I wrote about in the last post and which I presented in my CloudConnect keynote, but I added another angle, comparing the first attempt of creating a true public cloud, the Sun Grid, to the first successful attempt: Amazon Web Services.
The main difference between the two? You guessed it.
While the former (Sun) targeted the traditional IT customer, the CIO, Amazon went after developers, and that was the secret to their success.