Thinking Out Cloud

Ravello Launches the Cloud Application Hypervisor

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.

Related articles

Cloud Computing Adoption is Bottom-Up

Ravello Systems Raises $26M And Launches App Cloud Capsule

Ravello raises $26M to virtualize web development in prettier skies

Cloud app launch startup Ravello Systems raises $26M

Ravello and the Developer's Application Hypervisor Attract $26M in Funding

State of the Open Source Cloud

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.

Cloud Adoption Patterns - Citrix Synergy 2012 Keynote

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.

Regular readers of my blog know I write about this topic a lot (see recently Cloud Computing and SaaS Models Are About Bottom-Up Adoption and this post on the CloudSleuth Blog). In this talk, besides describing the phenomenon, I move a step further to discuss in more detail the implication of this adoption pattern to two groups: enterprise customers and SaaS/cloud vendors.

After the talk there will be a link to the video and slides. If you can't make it to Moscone on Thursday you can watch it live on the web from here:  http://live.citrixsynergy.com/sanfrancisco/

In the Game of Clouds, You Win Or You Die: CloudStack

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.

But the catch is this comes on the heels of Eucalyptus's announcement that it is aligning with the AWS APIs. Eucalyptus, just a reminder, is another open source cloud platform that has been vying for leadership among enterprise customers.

Game of Clouds

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.

More on Bottom-Up Adoption of Cloud Computing and SaaS

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.

For more, check out the full post here.

 

Cloud Computing and SaaS Models Are About Bottom-Up Adoption

Last week I gave a keynote presentation at the CloudConnect conference in Santa Clara. The title of the presentation was: "Surprise! Your Enterprise is Already Using the Public Cloud."

Regular readers of this blog (or those who work with me) know I go on about this a lot: In the enterprise, cloud computing services (IaaS, PaaS, SaaS) are being adopted bottom-up. In other words, by the rank & file (developers, IT admins, business folk) and not top-down with a big strategic decision by the CIO.

That's what the keynote was about and the title was addressing the CIO, who is the last to know about cloud computing adoption within his or her organization.

If you're interested in this topic you can watch the video of the presentation (you need to scroll down to get to it) on the CloudConnect web site. 

You can also read this very good summary of my talk by Rich Miller on Data Center Knowledge.

UPDATE [March 1, 2012]: Charles Babcock at Information Week wrote a nice piece about CloudConnect, including coverage of my keynote: 4 Cloud Computing Pain Points That Still Hurt.

And finally, here are the actual slides via SlideShare:

CloudConnect 2012: Surprise! Your enterprise is already using cloud computing

View more PowerPoint from gevaperry

 

BlazeMeter: Launching the JMeter Testing Cloud

Today, BlazeMeter is publicly launching its cloud load testing service and announcing its $1.2 million funding round led by YL Ventures. I joined the BlazeMeter board of directors back in July and we've been preparing for this launch since, so it's exciting seeing it all come together.

Like several other startups that I've been involved with, BlazeMeter leverages open source software in a cloud service, making web application development a whole lot easier.

Specifically, BlazeMeter uses Apache JMeter, the popular open source performance testing
framework, to create massive volumes of realistic browser simulations. BlazeMeter also allows current JMeter users, who have an existing set of JMeter scripts, to instantly load those scripts to the cloud and run them without any changes. Alternatively, folks can simply enter a URL, choose a pre-defined test scenario and run it instantly (with approporiate security measures when requesting very large stress volumes).

The beauty of what Alon Girmonsky, BlazeMeter founder and CEO, and his team did with it, is that although BlazeMeter is extremely easy to use, it is an enterprise-grade performance testing tool, both in terms of scalability and in terms of the comprehensiveness of the reports and analysis it provides.

In addition, BlazeMeter's pricing model is extremely attractive with a combination of usage-based pricing and subscriptions. And you can start running tests for free.

The company has a lot of great plans in store and I will have a lot more to say about it but for now, congrats to Alon, Daniela and the rest of the team! Alon is an incredible enterpreneur and I look forward to working with him on building BlazeMeter to a great company in the coming months and years.

 

Check out Alon's intro blog post, and don't forget to follow them on Twitter: @BlazeMeter for updates.

And see some additional coverage on the company today:

• The Register: Cose-probing, not Angry Birds, will define cloud's success by Matt Asay
• CNET: BlazeMeter raises funds for cloud-based load testing by Dave Rosenberg
• ITBusinessEdge: Changing the Economics of Load Testing via the Cloud by Mike Vizard

Public Vs. Private Cloud (yes, that old debate)

Hope everyone in the U.S. had a great Thanksgiving.

This is a quick follow up on my previous post, 10 Predictions About Cloud Computing, which received quite a bit of attention in the cloudosphere. Specifically, my second point there, which stated:

Public Rules: Internal clouds will be niche. In the long-run, Internal Clouds (clouds operated in a company's own data centers, aka "private clouds") don't make sense. The economies of scale, specialization (an aspect of economies of scale, really) and outsourcing benefits of public clouds are so overwhelming that it will not make sense for any one company to operate its own data centers. Sure, there need to be in place many security and isolation measures, and feel free to call them "private clouds" -- but they will be owned and operated by a few major public providers.

A coouple of weeks ago I had an interesting conversation with Marten Mickos, CEO of Eucalyptus (and of MySQL fame), a private cloud platform provider. I explained that I believe that Internal Private Clouds (i.e., clouds operated on a company's own servers), will become a niche in the long run.

Surprisingly, Marten did not disagree, but he made the following very good point. If we look a few years ahead, let's say 2015, the IaaS market in total will be a roughly $20 billion market according to estimates by IDC and others. If private IaaS is anywhere between 10% to 20% of that, certainly within the bounds of the definition of a "niche", we're still looking at a multi-billion dollar market. Enough to sustain the success of several startups and large vendors.

Point well taken.

The other follow up on this topic is a reference to my friend Nati Shalom's post on the topic. It's a good read with good arguments.

10 Predictions About Cloud Computing

Next week in Tel Aviv I'm going to participate in a panel about "the future of clouds", moderated by the legendary Yossi Vardi. In preperation, I wrote down a few of the concepts I've been thinking about for the past several years and I thought I would share them with my readers to get some feedback. Keep in mind these are long-term predicitions and trends (in no particular order).

1. PaaS Rules :IaaS becomes niche. In the long-run, IaaS doesn't make sense, except for a limited set of scenarios. All IaaS providers want to be PaaS when they grow up.


2. Public Rules: Internal clouds will be niche. In the long-run, Internal Clouds (clouds operated in a company's own data centers, aka "private clouds") don't make sense. The economies of scale, specialization (an aspect of economies of scale, really) and outsourcing benefits of public clouds are so overwhelming that it will not make sense for any one company to operate its own data centers. Sure, there need to be in place many security and isolation measures, and feel free to call them "private clouds" -- but they will be owned and operated by a few major public providers.


3. Specialized Clouds: There are many dimensions to an application: the pattern of its workload; the government regulations it must adhere to; the geographic access to it; the programmig language and framework it supports; the levels of security, performance and reliability it requires; and other more specialized requirements. It's not a one-size-fits-all world. At least, not always. There will be big generic clouds, and then, many specialized clouds. I've written about this in the past.


4. Government Regulation: The largest cloud providers will become nationally strategic infrastructure (like utilities, financials, telcos, airlines and shipping companies in the past). Given my "public rules" prediction above, cloud providers will become crucial infrastructure to the economy and the interests of their respective nations. They will become "too big to fail". Any change in their pricing will have a profound effect on the economy. And they will also hold the risk of a "cloud run" (similar to a "bank run", a sudden surge of demand they haven't anticipated. Not to mention the fact that they will maintain the sensitive data of consumers, corporations and government agencies. Any way you slice it, it spells regulation. But if history teaches us anything, this regulation will only come after "The Great Cloud Catastrophe" (use your imagination to figure out what that will look like). 


5. The Control vs. Freedom Debate: This sums up the story of cloud computing so far. Freedom is the catch-all phrase for drivers of cloud adoption (no upfront costs, on-demand, self-service, empowerment of the rank & file - e.g., developers), but control (or lack thereof) is the catch-all phrase of barriers to adoption by large enterprises. Every democratic country experiences this: there is sometimes a contradiction between the so-called sacred principles of rule-of-law and personal freedom. It's a matter of drawing the line -- and we're just in the beginning stages of understanding this issue when it comes to cloud computing. This debate will be with us for years to come and will shape the variety of enterprise cloud computing offerings.


6. Cloud Federations - While AWS has enjoyed tremendous international success, in any business that relies heavily on trust, such as IT, nothing beats a local brand. So people will flock to the cloud of their trusted national telco or big IT provider. But on the flip side, they will need to reach a global audience and will want servers around the world. As a result, we will see the formation of cloud federations, similar to what we see in airline alliances, such as Star, SkyTeam and Oneworld.


7. Financial Efficiency and Sophistication: Computing is a commodity, and every commodity ends up being traded, future-traded, brokered, arbitraged, speculated and manipulated with derivative instruments. The good: the market becomes very efficient. The bad: the market becomes complex and opaque. We are already seeing spot markets 


8. Cloud Standards: About two years ago there was a strong wave of interest and discussion about the need for cloud standards. I wrote then, and still believe, it is too soon. But it is also inevitable. We will, however, see multiple competing standards. At least one formal stanard specification from a standards body and several de facto standards from large commercial players such as Amazon and VMWare. 


9. The Ecosystem Wars: I've recently written about the importance of ecosystems in cloud computing. Success in building an ecosystem will be a determining factor in who wins and loses in the cloud. It is not just about the size and breadth of the ecosystem, but how well it all works together. In many ways, Amazon has done a poor job of this so far, but it has the one big compelling factor for an ecosystem: a very large install base. 


10. Horizontal and Vertical Consolidation: As with any industry, as cloud computing matures, it will consolidate. This will happen both horizontally, for example large IaaS players will roll-up regional and smaller IaaS and hosting providers, as well as vertically, for example IaaS providers will acquire cloud management system providers such as RightScale and enStratus.

I'd love to hear some feedback on these trends. Do you agree? Disagree? Have I left something out? Please let me know in the comments.

Java PaaS

[UPDATE (Feb 26, 2012): I added Jelastic and created another category I'm calling Private Java PaaS in which I put GigaSpaces Cloudify, Cumulogic and ActiveState Stackato.]

[UPDATE (Apr 12, 2012): Added DotCloud and rearranged the list a bit]

A couple of weeks ago during OracleWorld/JavaOne, Oracle announced its public cloud offering, including its Oracle Java Cloud Service -- the latest entrant into the increasingly crowded Java Platform-as-a-Service space. Despite being the most popular programming language, PaaS offerings for Java were not adopted as quickly as those for Ruby, Python and PHP.

Not surprisingly, however, once Java PaaS arrived at the scene, many of the big players now have PaaS offerings -- given Java's popularity in the enterprise. 

Here's the list I compiled (in alphabetical order). If I have overlooked anything, please let me know in the comments:

Public Java PaaS

• Amazon Elastic Beanstalk 
• CloudBees Run@Cloud
• DotCloud
• Google App Engine
• IBM SmartCloud Application Services
• Jelastic
• Microsoft Azure
• Oracle Java Cloud Service 
• OutSystems Agile Platform
• Red Hat OpenShift 
• Salesforce.com Heroku for Java
• VMWare CloudFoundry
• WSO2 StratosLive 

Private Java PaaS

• ActiveState Stackato
• Cumulogic 
• GigaSpaces Cloudify
• WSO2 Stratos

It should be noted that HP has a also made some noise about a Java PaaS but it hasn't launched yet.

Additional reading:

• Java PaaS Shootout: A technical comparison of Google App Engine, Amazon Elastic Beanstalk, and CloudBees RUN@Cloud