Thinking Out Cloud

Video: Cloud Databases Panel at GigaOm Structure

Without further ado, here's he video of the cloud database panel I moderated earlier today at GigaOm Structure:

gigaomstructure on livestream.com. Broadcast Live Free

 

VMWare's CloudFoundry.com Announcement

Today VMWare is about to make a big announcement about CloudFoundry.com. I'm writing this post before the actual announcement was made and while on the road, so more details will probably emerge later, but there is the gist of it:

VMWare is launching CloudFoundry.com. This is a VMWare owned and operated platform-as-a-Service. It's a big step in the OpenPaaS intiative they have been talking about for the past year: "Multiple Clouds, Multiple Frameworks, Multiple Services".

For those of you keeping track, CloudFoundry is the official name for the DevCloud and AppCloud services which have come out in various alpha and beta releases in the last few months.

The following diagram summarizes the basic idea behind CloudFoundry.com:

In other words, in addition to running your apps on VMWare's own PaaS service (CloudFoundry.com), VMWare will make this framework available to other cloud providers -- as well as for enterprises to run in-house as a private cloud (I'm told this will be in beta by the end of this year). In fact, they are going to open source CloudFoundry under an Apache license.

They're also going to support multiple frameworks, not just their own Java/Spring framework but Ruby, Node.JS and others (initally, several JVM-based frameworks). This concept is similar to the one from DotCloud, which I discussed in What's the Best Platform-as-a-Service.

On the services front, they are going to provide multiple services provided by VMWare itself, and eventually, open it up to the ecosystem for third-parties (similar to Heroku Add-Ons or Salesforce.com AppExchange).

Where it says in the diagram above "data service", for example, VMWare already has three offerings: MySQL (similar to Amazon RDS), MongoDB and Redis. For "message service" they will offer their own Rabbit MQ and other messaging services.

Finally, you'll notice there is reference to a "Downloadable 'Micro Cloud'" in the diagram. This is a free offering from VMWare that lets you run a CloudFoundry cloud on a single VM, which you can carry around on a USB memory stick or run on an Amazon Machine Image (which is what Michael from RightScale is going to demonstrate today). The idea behind the Micro Cloud is to appeal mostly to developers and let them easily do development in any physical location and seamlessly load their app to a CloudFoundry cloud when they are ready.

The Micro Cloud is one more aspect of this intiative that is intended to appeal to developers and encourage bottom-up adoption for the VMWare cloud. I've discussed the idea of developers being the driving force of cloud adoption before, and you can read more about it in this post. 

All in all, a very smart, if not unexpected, move by VMWare. But it remains to be seen how VMWare will handle the inevitable conflict between being a cloud provider and hoping to be the provider of infrastructure software for other providers. Case in point: VMForce, the joint offering announced by VMWare and Salesforce.com several months ago.

Officially, both companies say everything is on track for the VMForce offering, and VMWare says VMForce is "powered by CloudFoundry". But as I am quoted in this TechTarget article, the companies were already on a collision course, especially after Salesforce.com's Heroku acquisition.

From talking to some of the folks at VMWare, it's clear that they too believe that the future of computing is PaaS -- something I believe strongly in. It will be interesting to see how they execute on this grand vision.

Video Interview on Zenoss's Cloud Management Blog

Josh Duncan at Zenoss, which makes cloud monitoring and management software, interviewed me over Skype for their Cloud Management Blog. We covered a pretty wide range of cloud-related topics and Josh split up the video interview into two parts.

The first part is now up on their blog here. (by the time I post this blog the second part may be up as well).

To quote from Josh:

Here’s what we cover in part 1 of Thinking Out Cloud with Geva Perry:

• “It’s happening now” – what are all the challenges and strategies to address
• The business case for the Cloud – from cost savings to business agility
• The operational challenges of Cloud computing
• “The CIO doesn’t matter” - democratization of IT (link to Forbes article)

See the video.

UPDATE: Part 2 is also avilable now at: http://blog.zenoss.com/2011/03/thinking-out-cloud-with-geva-perry-part-2/ 

What's the best Platform-as-a-Service?

Recently there has been a surge of interest in PaaS. It is partly fueled by the maturing of the various offerings and partly by industry developments, such as the $250 million acquisition of Heroku by Salesorce.com (disclosure: I was an advisor to Heroku) and Amazon Web Services releasing Elastic Beanstalk and CloudFormation. Case in point: just yesterday, PaaS player DotCloud announced a $10 million funding round from top-tier funds Benchmark and Trinity.

Not a lot of information is provided on the DotCloud web site and announcement -- they are still in private beta -- but one thing is clear: they are positioning themselves against Heroku.

In the announcement, DotCloud bills itself as a "2nd-generation PaaS" (thank god not "PaaS 2.0") and the news articles about it imply that Heroku is first generation.

Here's a quote from TechCrunch:

DotCloud’s appeal is in its flexibility. Unlike the first generation Heroku and its clones, DotCloud lets developers to build and deploy their applications rapidly by allowing them to customize components and supporting multiple languages and tools.

This is also the positioning on DotCloud's home page, where they state in one of their four key benefits points (italics mine):

Not just for toy apps.

Anyone can deploy "Hello World" in 30 seconds. But you're building a product. You need the freedom to mix and match languages, frameworks and databases as you see fit. DotCloud supports that. Our competitors don't.

I have no doubt that DotCloud has a great concept, technology and team -- if only because smart nvestors like Peter Fenton, Ron Conway, Chris Sacca, Dan Scholnick and others put so much money in them -- but they are missing the point on this one.

The approach of giving a more flexible platform -- one which allows you to "mix and match languages, frameworks and databases" and in which you can "pop the hood", as DotCloud seems fond of saying -- that idea is not a more evolved or better generation of PaaS; it's simply a different one.  

It's not a new idea in tech and software that there is a trade-off between flexibility and ease-of-use; between catering to the 20% power users and the 80% everyday users; between general purpose platforms and specialized platforms. PaaS is no different. 

More than a year and a half ago I had written about the need for a variety of specialized clouds (and specifically PaaS clouds), which are different from each other on a number of dimensions, which included things like: framework support, usability, application use case, domain (i.e., functionality) and more. 

A multi-language PaaS is not superior to a single-language PaaS. There must be a trade-off in the technology itself, and also a different go-to-market approach.

From a technology perspective, a very focused platform, such as Heroku in the Ruby space, has many advantages. Everything is tightly integrated, pre-installed and pre-configured. This shields the user from much of the unwanted "under the hood" and repetitive tasks. It also makes sure that -- accepting the limitation of ruby only -- the platform can provide much richer functionality out-of-the-box and much shorter development and deployment cycles.

The DotCloud approach, on the other hand, gives more flexibility and would serve a more complex class of applications that indeed use multiple languages and frameworks. But the trade-off is more complexity, both in their own development efforts as well as for their users. Which brings me to my second point about go-to-market strategy.

As a general rule, startups need to focus on a well-segmented community of customers. For Heroku, the intense focus on ruby developers was a major contributor to their success (for more on this see my post from June 2009 Ruby Developers: The Cloud Generation). It allowed them to keep their message very focused and direct all of their marketing efforts (and budget) on a very specific audience.

I am not sure what DotCloud's market segment is yet. It's not very clear from their home page, but given that they are in private beta and just recently received funding, that's okay. They will, however, eventually have to find their focus. Displaying twenty logos of languages and frameworks on their home page is nice, but may also be confusing and off-putting to many developers.

Developers (and non-developers) now have a large and growing range of platform-as-a-service offerings to choose from. Just to give you a taste of the variety here are some PaaS examples (in no particular order), each one focused on different combinations of technology platforms and/or types of users and applications: 

Amazon Elastic Beanstalk, Microsoft Azure, Google App Engine, Heroku, VMForce, Makara/Redhat, Force.com, VMWare Cloud OS (aka AppCloud/DevCloud), DotCloud, Cloudbees, PHPFog, Engine Yard, Application Craft, Morph Labs mCloud On-Demand, OrangeScape, WorkXpress, Wix.

Although we will see more players emerge in the space, not all of them will be winners. That said, to answer the question in the title of this blog post: What's the best platform-as-a-service? It depends on who you are and what you're trying to do.

[Stay tune for more on this topic and the players mentioned above].

Two Computerworld Articles on Cloud Computing

Yesterday a couple of Computerworld pieces in which I'm quoted came out. They are both authored by Beth Schultz.

The first is about Cloudonomics, or how can enterprises figure out the potential cost-savings and other financial effects of cloud computing. My basic take was that it's difficult to measure the exact financial impact of cloud computing because one of it's major benefits is business agility. See the full story.

The second is about the plethora of Cloud Services available to enterprises and how to choose among them. My take on this one was that it's not a one-size fits all game and organizations will need different tools for different tasks. Read the full story.

Evaluating Cloud Computing Services: Criteria to consider

Today TechTarget published a piece I wrote for them about criteria to consider when selecting a cloud provider. Here's the opening paragraph:

Selecting a cloud computing provider is becoming increasingly complex. As cloud environments mature, many cloud providers attempt to differentiate themselves by focusing on specific aspects of their offerings, such as technology stacks or service-level agreements (SLAs). In short, not all cloud providers are created equal.

At the same time, enterprises are beginning to rely on cloud providers for hosting mission-critical applications, which raises the stakes for selecting the right cloud service. So how do organizations navigate this multifarious landscape? Below you'll find a few key factors for evaluating services as well as some resources to use.

I then go on to cover six key areas:

• Performance
• Technology stack
• SLAs and Reliability
• APIs: lock-in, community and ecosystem
• Security and compliance
• Cost

Please read the full article here.

Death of an (Enterprise) Salesman

Zack Urlocker wrote a nice piece on GigaOm about the Consumerization of IT, a topic that's near and dear to my heart, as readers of this blog probably know. The main focus of Zack's post was that enterprise software needs to be more simple and with less features, similar to consumer software and electronics.

But he also writes:

In a recent presentation, Lew Cirne, CEO of application performance management vendor New Relic, revealed the company now has 5,000 customers and just one sales rep. That’s astonishing.

Again, as regular readers of my blog probably know, this is a pet peeve of mine, which I write about often (for example, here, here and here). And it's not that astonishing if you know what's going on at many cloud and SaaS companies. You need only look at companies such as Amazon Web Services, Heroku and Twilio, to name a few,  who are enjoying similar or greater efficiencies in their sales and marketing efforts, and whose customers include large enterprises, not just SMBs (as do New Relic customers).

As full disclosure, I've worked as an advisor with New Relic, Twilio and Heroku, and am now working with companies such as Xeround, Sauce Labs and others (including large publicly-traded telco and hosting vendors), and we are successfully building similarly effective no-touch sales models.

The reason I mention this is because despite the fact that there is a now a growing list of companies wh are having success with this model in the B2B/Enterprise space, there is still much resistance to this model by boards and investors and it's a shame because they are giving their portfolio companies bad advice.

More than a year ago I wrote Marketing Cloud Computing: Uncharted Territories, but these territories are now a lot less uncharted, and we're starting to see a playbook emerge for this model. I am going to write more about this playbook in coming posts.

 

QCon San Francisco 2010 Cloud Track

Back in July I wrote that I am hosting the QCon San Francisco Cloud Computing track this year and made an informal call for speakers.

Well, I wanted to update on the awesome line-up of speakers that we have for this event and offer my blog readers a $100 discount on registering to the conference, which takes place November 1-5 at the Westin San Francisco. The cloud track takes place all in one day on Friday, November 5.

To receive the $100 discount, sign up here with registration code PERR100.

Here's the agenda with detailed abstracts below. Hope to see you all there!

Track:	Real Life Cloud Architectures
Host	Geva Perry
10:35	Get Satisfaction uses Ruby on Rails and cloud computing platform to achieve scalability and reliability Thor Muller
12:05	Implementing private clouds Andrew Clay Shafer
14:05	Netflix’s Transition to High-Availability Storage Systems Siddharth Anand
15:35	Out of This World Cloud Computing John L. Callas & Khawaja Shams
16:50	Panel: Data in the Cloud Amr Awadallah & Roger Bodamer & Simon Guest &Damien Katz & Razi Sharir

 

Track: Real Life Cloud Architectures

Host: Geva Perry

This track covers how a variety of applications are using Cloud computing today, with a focus on mature organizationss who have merged some aspect of cloud into their offerings rather than new/small companies which started out in the cloud. The aim of this track is to provide concrete takeaways for how to start using cloud computing in-house when you have an existing application to work with.

Presentation: "Get Satisfaction uses Ruby on Rails and cloud computing platform to achieve scalability and reliability"

Track: Real Life Cloud Architectures

Speaker: Thor Muller, CTO and Co-Founder, Get Satisfaction

Time: Friday 10:35 - 11:35

Location: Olympic Room

Abstract:

Get Satisfaction provides a customer community platform to help companies respond to customers’ questions, feedback, concerns, and praise – no matter where the conversation is taking place across the social web, including Twitter, Facebook, and blogs. With 5 million customers per month generating millions of monthly page views, scalability is a priority for Get Satisfaction. The company needed a cloud platform that would be able to keep up with its rapid growth rate, reduce the time and expense of infrastructure management, and improve upon slow deployment times that they had experienced with a dedicated infrastructure provider. Deploying the Get Satisfaction application to the cloud – not directly investing in hardware or heavy-duty architecture on the system side – allows the company to right-size its investment for growth.

Get Satisfaction chose to develop their applications with Ruby on Rails and has realized several benefits from using the fast-growing language and framework: agile methodology, test-driven, behavioral methodology, better code, reduced risk, and more performance visibility for the business. As a fast-growing start-up, Get Satisfaction wants a Platform-as-a-Service (PaaS) provider who would grow with them, while providing expert Rails advice. Get Satisfaction has been running on Engine Yard AppCloud since January 2010 – realizing the benefits of affordability, ease of deployment, higher performance and scalability, and 24/7 support.

Presentation: "Implementing private clouds"

Track: Real Life Cloud Architectures

Speaker: Andrew Clay Shafer, Cloud Building DevOps Guy, CloudScaling

Time: Friday 12:05 - 13:05

Location: Olympic Room

Abstract: On-demand virtualization is just the start in your private cloud journey.  Scalability and self-service mean building true horizontal cloud services that can be reused by all your enterprise applications.  Think GoogleFS or BigTable, Amazon SQS or RDS.  What do you need for a foundational baseline?  What services do you need on top of that baseline?  How do you communicate it to your enterprise IT department and as a worst case build it yourself?  We'll lay out what a complete private cloud ecosystem looks like and explore various options in a vendor neutral manner, beginning with virtualization and ending with PaaS-like capabilities deployable today.

Presentation: "Netflix’s Transition to High-Availability Storage Systems"

Track: Real Life Cloud Architectures

Speaker: Siddharth "Sid" Anand, Netflix Cloud Systems Team

Time: Friday 14:05 - 15:05

Location: Olympic Room

Abstract: The CAP Theorem states that it is possible to optimize for any 2 of Consistency, Availability, and Network Partition Tolerance, but not all three. Though presented by Eric Brewer in 2000 and proved in 2002, the CAP Theorem has only recently gained significant awareness and that primarily among engineers working on high-traffic applications. With spreading awareness of the CAP theorem, there has been a proliferation of development on AP (a.k.a. Available-Network Partition Tolerant) systems – systems that offer weaker consistency guarantees for higher availability and network partition tolerance. Much of the interest in these AP systems is in the social networking and web entertainment space where eventual consistency issues are relatively easy to mask. Netflix is one company that is embracing AP systems. This talk details Netflix’s transition to AWS SimpleDB and S3, examples of AP storage systems.

Presentation: "Out of This World Cloud Computing"

Track: Real Life Cloud Architectures

Speakers: Khawaja Shams, Software Engineer, NASA Jet Propulsion Laboratory & John L. Callas, Project Manager of NASA's Mars Exploration Rover project

Time: Friday 15:35 - 16:35

Location: Olympic Room

Abstract:

The Mars Exploration Rovers (MER), originally designed for only 90 days of operation, have been roving the surface of Mars for over six years. They have trekked tens of kilometers, well beyond the 1000-meter design capability and sent back over a quarter million pictures from the surface of Mars. The infrequent communication links, once or twice a day, present operations challenges, but they also epitomize bursty utilization of computational resources. The planning cycles are often as short as 4 hours, where even an hour of data processing consumes precious operational schedule. Lastly, with a global community of scientists, distributing content efficiently to the entire team is essential in conducting seamless remote operations.

This session will present MER operations as an excellent candidate for cloud computing. MER is the first NASA mission to effectively utilize a public cloud computing infrastructure and realize its benefits. We will discuss the cloud applications we have built for storage, content delivery, searching, and data processing. Come learn how this great mission of exploration has been able to use cloud services in production to solve mission critical problems, while saving money and streamlining operations. We will also cover challenges associated with security and privacy, as well as designing our services to maximize uptime and minimizing cost.

Panel: "Data in the Cloud"

Track: Real Life Cloud Architectures

Moderator: Geva Perry

Panelists:

• Amr Awadallah, CTO, Cloudera
• Roger Bodamer, SVP Products, 10Gen/MongoDB
• Simon Guest, Microsoft Senior Director of Technical Strategy in the Developer and Platform Evangelism
• Damien Katz, Creator of CouchDB and CouchOne Founder
• Razi Sharir, CEO, Xeround - SQL cloud database

Time: Friday 16:50 - 17:50

Location: Olympic Room

Abstract: Scaling out the data tier has always been a challenge, but it is exacerbated in the cloud. By nature, cloud computing environments are dynamic and elastic. That, coupled with the fact that today's applications need to handle web-scale data volumes and transactions, makes the traditional approaches to data management unfeasible. This panel will discuss the unique challenges of managing data in the cloud -- and the solutions from a variety of perspectives: database-as-a-service, NoSQL, cloud-specialized relational databases and more.

 

AutoCAD in the Cloud Launches Today

Although it announced it a few weeks ago, today Autodesk is launching its cloud service, AutoCADWS, and its iOS apps (iPod Touch, iPhone, iPad) for storing, editing and sharing Autocad DWG files in the cloud.  

It's a nice follow up to my Oh, SaaS, is there anything you can't do? post from a couple of months ago. This isn't quite a true SaaS application (yet), but it shows, once again, how cloud computing can apply to a wide variety of applications and how companies are re-thinking the delivery models of their products.

At the time of me writing this post the link to the cloud service, AutoCAD WS, wasn't live yet, but according to the company it should be launched today. My wife is an interior architect and I'm sure she's going to want to play around with this new service. The iPhone and iPad apps are already available on the iTunes store (for free).

I am looking forward to speaking to some of the Autodesk executives to better understand what their plans are for the cloud, as I understand they have a lot more goodies in store.

UPDATE: The AutoCAD WS site is now live.

Shopping the Cloud: Pricing (or Apples and Oranges in the Cloud)

In Shopping the Cloud: Performance Benchmarks I listed a number of services and reports that compare cloud provider performance results, but the truth is that in computing (cloud included) you can throw money at almost any performance and scale problem. It doesn't make any sense, therefore, to talk about performance alone, you want to compare price/performance.

But here's the rub: it is becoming increasingly difficult to compare the pricing of the various cloud providers.

Problem #1: Cloud providers use non-standard, obfuscated terminology

About a year and a half ago I wrote What Are Amazon EC2 Compute Units? in which I raised the issue of how difficult it is to know what it is you are actually getting for what you are paying in the cloud. Other vendors use their own terminology, such as Heroku's Dynos. I'm not just picking on these two, everyone has their own system.

Problem #2: Cloud providers use wildly varying pricing schemes

In addition, the pricing schemes by the various vendors include different components. Take storage as the simplest example, which clearly illustrates the point. Here's a screenshot from the Rackspace Cloud Files pricing page:

 

It is fairly straightforward, but also contains many elements that are extremely difficult to project (especially for a new application), such as the Bandwidth and Request Pricing. That's OK - you have to make some assumptions. 

But here's my main point -- now compare it to Amazon S3 pricing:

Good luck with that.

Problem #3: Not all cloud offerings are created equal

To make things worse, not all cloud storage services were made equal. They have different features, different SLAs, varying levels of API richness, ease-of-use, compliance and on and on.

Problem #4: Cloud computing pricing is fluctuating rapidly

Another big problem with dealing with pricing is that the market is very dynamic and prices change rapidly. Fortunately, most of the movement right now is downwards, due to the increased competitiveness (especially in the IaaS space) and thanks to vendors benefiting from economies of scale and increased efficiency due to innovation.

Andrew Shafer from CloudScaling wrote a blog post a couple of weeks ago in which he shows how Amazon pricing is constantly shrinking. Check out this graphic he created:

What to do about it?

So what do you do in such a complex landscape? There seems to be no escape from creating a test application and running it on multiple services to see where the cost comes out. Then again, that may turn out to be a very time-consuming and expensive effort that may not be worth it -- at least not initially. So you should be prepared to have to move your app across cloud providers if and when the costs become prohibitive (which I am seeing happening to more and more companies).

Hopefully, the cloud benchmark services will also start paying attention to pricing and provide a comparison of price/performance and not just performance.