Lew’s background spans more than 20 years during which he has been instrumental in driving several major  technology changes, including: AI and massively parallel systems, developer adoption of Java, Salesforce.com’s AppExchange,  and most recently, Cloud Computing.    According to Lew, “At Thinking Machines, in the early 1990′s, we were building massively parallel machines using thousands of individual processors.   At Sun, we drove the evolution of the web with Java and networking, often using the tagline, ‘The Network is the Computer’.   In this next phase, it’s becoming clear that the ‘Cloud is the Computer’  and this promises to be just as disruptive.”

Cloudscaling CEO, Randy Bias, and Lew Tucker both share a long-term interest in the design and architecture of large, scalable systems.  As former VP Technology Strategy of GoGrid, Randy was responsible for building out one of the most complete infrastructure services in the cloud.  As CTO of Cloud Computing for Sun Microsystems, Lew was responsible for the architecture and API for Sun Cloud.

Lew’s joining the advisory board continues to build up the Cloudscaling team’s unique set of resources.  “If you want to build significant clouds, you have to have the right team.” said Randy Bias.  We’re the only cloud engineering services team I know of that can point to not one, but many, large scale cloud environments they have built.”

About Cloudscaling

Cloudscaling is the leading cloud computing engineering services firm. We provide strategy, design and implementation to build cutting edge clouds. Located in San Francisco, the company was founded by experts who have built some of the largest public and private clouds operating today. Visit cloudscaling.com to read the blog and follow the team on twitter.com/cloudscaling.

Contact:
Pat Sharp
pat@cloudscaling.com
725 Cool Springs Blvd., Ste. 600
Franklin, TN
USA
Ph: +1 (615) 732-6192

###

Spectrum of Cloud Computing Providers

There will soon be two major paths for cloud computing providers: commodity and premium.  If you read my series, Cloud Futures, you’ll know that I broke down cloud service providers into three major categories: service clouds, consumer clouds (previously ‘commodity’)[1], and focused clouds. In retrospect I realize now that there are possibly four, not three major categories. The missing category is premium enterprise clouds. Previously I had lumped these under focused clouds, but I now realize that, in fact, there are likely to be so many of these that they deserve their own category. I’ll go even further and suggest that in terms of markets targeted, there will really only be two ends of a spectrum: enterprise and non-enterprise.

Most clouds will fit towards one of this spectrum or the other. In essence, you’re targeting small businesses (startups, SaaS providers, and SMBs) or you are targeting larger businesses (SME or Fortune 2000). The former are extremely cost conscious while the latter may have a number of other equally important drivers, such as security (e.g. VPN access), high availability (HA), SLAs, application portability without modification[2] and similar. Clearly large enterprises will consume services at both ends of the spectrum, but they will have many use cases (mostly ‘production’) that can only be serviced by a premium service running VMware’s forthcoming vCloud product.

This means we will have a large bifurcation in the cloud computing space with two very different kinds of solutions. Clouds will either target commodity customers or premium customers.  Very few clouds will actually fit in the middle of this spectrum initially, although I expect providers on both sides will grow towards the middle.  In quite a few cases (AT&T and Rackspace come to mind) cloud providers will build two offerings at both ends of the spectrum, but we haven’t seen this quite yet.

Premium vs. Commodity

Ultimately, commodity clouds will be forced via pricing pressure to continue to drive down capital expenses and operating costs.  As we can already see in the public cloud space, providers have largely standardized on the Xen open source hypervisor.  This is the de facto standard because it is free.  In contrast, premium enterprise clouds will necessarily spend more on their infrastructure to provide advanced features like HA.  Their pricing will reflect this, but it also means they will use VMware’s products and hence have unique opportunities for integrating with internal clouds at large enterprises (more on this below).

This table summarizes the differences.

Enterprise Clouds Are Already Here

If you were paying close attention this year, you’ll have noticed that both Savvis and Terremark are working on or have delivered enterprise cloud offerings.  There are many  more on the way.  These providers are delivering VMware-based platforms specifically for enterprise customers and pricing reflects that[3].  Terremark even labels itself ‘The Enterprise Cloud’.  I had hoped to release a full review of Terremark, but due to time constraints haven’t been able to complete it.  If you click on the screenshot to the right it will take you to a set of Flickr photos that are an extensive tour of the Terremark Enterprise Cloud product.

What’s most interesting about this is that two major players have entered into this space and at the same time VMware’s vCloud is unreleased.  Nor are there any other shrink-wrapped software packages for building a cloud based on VMware.

VMware’s Dominant Position for Building Internal Clouds

But why VMware?  What’s so important about it?  For those of you who may not be aware, VMware’s enterprise-class hypervisor (ESX) is the de facto standard inside the enterprise, in much the same way the Cisco routers & switches are a standard.  This means that as enterprises move towards building internal clouds (an inevitability), they will be more likely to build clouds based on VMware’s ESX, which they are already comfortable with.

A-ha! Surely there is a startup or major player who has already delivered a software offering that allows enterprises to build their own internal clouds?

No. There is no credible contender to VMware’s crown.  Even though they did not see cloud computing coming, even though they are a large organization and slow to move, there is still not a single credible contender with a released product that manages the VMware ESX hypervisor and allows you to build a real self-service internal cloud.  Nada.  Zip.  Zilch.

There are some prospects like Platform’s ISF[4] that could be contenders, but by the time they are released in the wild, VMware’s vCloud will also be released.  The window of opportunity for making significant inroads into the enterprise is closing quickly[5].  Once VMware’s vCloud is released, who will risk averse IT managers and CIOs in enterprises go to?  A new player or someone already trusted and embedded like VMware?  There is no doubt.  They will largely select vCloud unless VMware fails to execute.

Can VMware Fail to Execute?
Is it possible for VMware to fail to execute in it’s sweet spot?  It’s area of expertise?  Yes.  Is it likely?  No.  If you look at the DNA of the business they already have the kinds of talent necessary for building a strong product in their acquisition of Akimbi, the folks upon whom the VMware Lab Manager product was built.  That team already knows how to build a self-service portal and a large scale VM deployment system including scheduler as these were integral for the lab manager product.

In other words, the writing is on the wall.

The Power of Internal + External Clouds
For many smaller business, moving everything to the cloud will always be a very compelling solution, but for the enterprise it will never be acceptable.  For various reasons (regulatory, political, legal, and others) enterprises must maintain a certain amount of infrastructure.  Also, I’ve heard fairly compelling arguments that large enterprises have sufficient scale to build and operate their own clouds at a cost advantage to external clouds.  Regardless, some capacity will reside outside of the firewall.

The usage of external clouds will largely be dictated by use case and in order for enterprises to derive maximum value from both internal and external clouds they will want a single internal portal that manages both.  They will want minimal friction for internal customers to be able to pick the best cloud for the job/cost.  It will also be important to allow some amount of portability (moving VMs and their workloads across the firewall).

While this doesn’t require a VMware hypervisor on both sides of the firewall, it will be greatly facilitated if that is the case.  Tools written against the vCloud API will likely work with vCloud-based external clouds without modification.  There is simply far too much synergy possible once both internal and external clouds are based on the same cloud platform.

Summary
There will be two paths for clouds: premium & commodity.  Premium clouds will focus on the enterprise and delivering value they are concerned about.  Commodity clouds will largely be forced to compete on pricing and features irrelevant to the enterprise.  VMware’s vCloud will be the dominant player behind the firewall because there is no credible contender.  The synergistic effects of internal & external clouds being based on the same vendor’s software will provide powerful and compelling reasons for enterprises to adopt those external clouds.  Enterprises will use commodity clouds, but mostly for batch processing and non-production workloads that are pricing sensitive.  The bulk of enterprise cloud spending will be on vCloud-based public cloud providers.

1. The Open Cloud is Coming
2. Defining Infrastructure Clouds
3. EngineYard uses Chef, a Puppet Alternative
4. Virtual, Cloud, Datacenters
5. My GoGrid Status (an explanation of my departure from GoGrid)

I hope that if you are a newcomer, these give you a little more flavor and insight into what this blog is about.

Best,

–Randy Bias

This slipped by me, but looks like GoGrid made the famous Gartner ‘magic quadrant’ for Web Hosting and Hosted Cloud Infrastructure Services (On Demand).[1]  You can see the quadrant here to the right.

It’s great that GoGrid is listed in the visionary quadrant.  When I joined GoGrid in the office of the CTO, one of my explicit goals was to increase the view of GoGrid as a thought leader and visionary in the cloud computing space.  It gratifies me to see the recognition by Gartner.

For those of you who may have missed it, I left GoGrid recently and am currently an independent Cloud Strategist.

At the moment my client list is short, but impressive:

• VMware (#1 Virtualization system; working with their new Cloud business unit)
• GoGrid / ServePath (#2 or #3 IaaS provider in U.S. depending on how you count)
• EngineYard (#1 Ruby on Rails PaaS provider)
• Cloud Central (first IaaS provider in Australia; pre-launch)

Find out how I can help you with your cloud strategy in my one page briefing (PDF).

Having been ‘on the inside’, I know that GoGrid leads this space with their technology and their commitment to be the world’s #1 ‘cloudcenter‘ (still one of my most widely read GoGrid blog posts).

New Directions
There are several exciting opportunities I’m pursuing.  Standby to learn more.  :)  I’m also exploring ways to work directly with cloud customers and cloud providers. My unique knowledge of cloud technology enables me to offer my expertise to those looking for new ways to exploit the cloud, ‘onboard’ to the cloud, or build their own cloud.

You can find out more in my one page consulting brief downloadable here or to the top right of my blog. 

More good news and blog posts coming your way soon.

— Randy L. Bias, Cloud Strategist

GoGrid — Scaling Your Internet Business

Mitch Denny, a member of the GoGrid developer community, stepped up and took over maintenance of the GoGrid .NET SDK.  He’s just released a new version that updates to support the latest GoGrid API features including jobs and pagination.  You can also find Mitch on twitter.

Thanks Mitch!

Where ever you stand on ‘free markets’, there is no doubt that market forces play a major role in technology adoption.  In this case it means that not only can cloud computing be open.  It will be open because of market pressures.

Open Source and Market Forces
Without a doubt open source has benefited from market forces, becoming widely embraced.  For many (e.g. startups) the cost economics were too compelling.  Others (e.g. Google) reaped both the benefits of cost economics and the openness of the source code itself.  Certainly even those companies who embraced open source for the ability to see the source code first and foremost were looking at it with an eye towards saving money elsewhere, not from ideology.

Market forces that drive open source are the same as those that drove the opening of Internet walled gardens.

Walled Gardens, Market Forces, & Democracy
The openness of the Internet is certainly not in doubt.  Despite ongoing attempts by providers to profit by selling a better quality of service to those with bigger pocket books, Net Neutrality advocates regularly stymie their efforts[1].  When large publishing sites and newspapers have closed off their access they are largely scorned by the individual Internet user.  Back in the day, we used to paraphrase John Gilmore saying: ‘the Internet routes around censorship, problems, or politics.’  Meaning that, like the network protocol (TCP/IP) upon which it is based, the Internet is largely immune to efforts to disrupt it.

Why is this and what are market forces really?  ‘Market forces’ are the effect of people voting with their wallets.  People want open, cheap, and accessible on the Internet.  It’s been proven over and over countless times now.  Internet-based walled gardens will eventually fail with few exceptions.[2]

Cloud computing is no exception.  It’s got to be open, cheap, and accessible.

Opening of the Cloud
It’s very easy to see the current advent of cloud computing as being closed.  It is.  Historically most new Internet markets and technologies are closed until they mature.  The cycle can be described as:

1. A new Internet market/technology starts as closed and proprietary
2. Under pressure in a high growth market it begins to open
3. When market or technology matures many of the technologies are ‘open’ or ‘open enough’

Consider for a moment the now standard x86 processor instruction set, the SQL relational database interface, or the availability of APIs for just about every Google service.

Already there are the inklings of people working on concrete standards including ones for cloud computing infrastructure (Open Cloud Computing Interface (OCCI) a working group of the Open Grid Foundation (OGF)) and the portability of virtual machines (Open Virtualization Format (OVF)).  Once these standards are finished and embraced by even a few clouds many customers will opt to choose cloud providers that use them, avoiding vendor lock-in and give choice to customers.

Clouds are not Software
Being ‘open’ means something different for Internet-based services, such as cloud computing, than it does for software.  The essence of cloud computing is that you are renting resources from someone else.  This is powerful and provides tremendous advantages to the customer.  It’s also no different now than it was when it was calling ‘hosting‘.  The very nature of this kind of offering is that customers won’t have 100% control.  Someone has to run and maintain the system that you are renting in a professional fashion and retains control … up to a point.

Open Cloud Services?
Where are the points of ‘openness’ for Platform-as-a-Service (PaaS) and Infrastructure-as-a-Service (IaaS) businesses?  It’s fairly simple to assess this.  What is ‘open enough’ to allow you to take your application and move it from one PaaS or IaaS provider to another?  Some obvious points come to mind:

• Data & code portability
• Widely available programming languages and libraries
• Interoperability through widely embraced and supported APIs
• You ‘own’ it and can take it with you

Where ‘it‘ is your web application (PaaS), virtual server (IaaS), and data.  Vendor lock-in comes from disallowing you to leave and take your data, tools, or software with you.

Platform Lock-in
If we were to measure PaaS providers like Salesforce.com’s Force.com, Google App Engine (GAE), and Heroku they might be graded as follows:

• Force.com:  ‘D-’
• GAE: ‘C’
• Heroku: ‘A’

Force.com creates extreme lock-in via its closed proprietary language Apex that is unavailable outside their service.  GAE a bit less so as it supports widely available languages such as Java and Python.  On the flip side, in order to leverage GAE in a meaningful manner you must use Google proprietary libraries and technology to access BigTable instead of a traditional database.  Heroku comes out smelling like roses allowing you to use a widely available language like Ruby with little or no modifications.

Infrastructure Lock-in
Unlike PaaS providers, infrastructure services do not dictate languages or libraries, meaning that there is even less areas for them to lock you in.  By providing a basic building block of a ‘virtual machine image’ IaaS providers allow customers to ‘own’ and ‘control’ their servers.  Even now, the 800lb gorilla of cloud computing, Amazon Web Services, allows you to download their Amazon Machine Images (AMI).  You can download one now and run it on your own infrastructure.  You can take it with you.

The primary way in which infrastructure providers create lock-in is twofold.  First  by using a proprietary API that is not widely adopted and second by disallowing portability of virtual machine images.

Once tools are written for a closed API you’ll get little or no reuse of those tools.  This is why folks like GoGrid, my employer, and our competitors are actively working together to create an open API for controlling IaaS systems.  As part of that discussion we’re also actively working on supporting standards like OVF that allow virtual machine image portability across hypervisors and clouds.[3]

If you were to grade a bunch of us today, however, it wouldn’t be bad, but it could be better:

• VMware vCloud/vSphere: ‘C-’
• GoGrid: ‘C-’
• Amazon: ‘C-’

You might score these differently, but for me VMware gets marks for widely adopted APIs, but demerits for dragging their feet on standards they helped create like OVF and being missing from conversations like the OCCI effort.  GoGrid has an open license on our API, but doesn’t allow downloading or uploading virtual machine images [4].  Amazon has widely adopted APIs, but under ambiguous licenses and quite a few proprietary technologies (e.g. S3, SimpleDB, and SQS) that encourage lock-in.

The good news is that standards for cloud interoperability (OCCI) and virtual machine portability (OVF) are either here or imminent and many providers are planning to adopt them.

Summing Up ‘Open Services’
It means something different to be an ‘Open Service’.  We all want to be able to export and move our applications from Google App Engine or Force.com when we leave.  We also want to take the virtual machine images that we build on Amazon or GoGrid and move them elsewhere.  Today we  can, but it’s just a nuisance.  New standards will make interoperability and portability easy and convenient. If anything, cloud computing is becoming ‘open’ faster than anything before it.

Ignore the naysayers.  Customers want choice and they will have it.  Choice is driven by open standards, cheap resources, and easy ‘self-service’ access.

The ‘market’ demands it. 

George Reese‘s new book, Cloud Application Architectures, is finally out.  I’m probably biased, but I think it’s pretty good as a handbook for building web applications on Amazon’s EC2. I contributed an appendix that describes GoGrid and it’s differences to EC2.  There is also a good interview on CloudAve with George Reese on the book.

Highly recommended as a reference manual.  

No it isn’t.  There are important differences and I’ll explain them and why it matters.

Defining Infrastructure Delivery Models
There are really three emerging infrastructure delivery models: datacenters, cloudcenters, and virtual datacenters.  Datacenters have been around forever, but cloudcenters and virtual datacenters are largely brand new options created by the convergence of computer virtualization, automation, and the on-demand business model.  These delivery models are not exclusive, but rather decreasing subsets of functionality that represent differing levels of virtualization and customer viewpoints.  For example, a datacenter can deliver cloudcenter and virtual datacenter services.  Cloudcenters can deliver virtual datacenters.  A virtual datacenter cannot ever provide traditional datacenter functionality however.  The following diagram shows the relationships:

Datacenters represent the most control, but largest expense.  Building and running your own datacenter or taking down co-location space from a datacenter provider and building out your own infrastructure on top of it is not only capital intensive, but requires a certain level of sunk cost in terms of operational expenses (OpEx) to be cost-effective.  This is my main test:  Do you already have the specialists (network, systems, storage, and facilities) employed? Do you need 1,000+ servers?  Then a traditional datacenter is probably right for you.

Cloudcenters represent the next step up, providing significant control, similar to datacenters, but provided in an on-demand delivery model where your only costs are monthly recurring fees.  Usually you have no need for many of the specialists you would normally need (e.g. network and storage).  Cloudcenters also provide many hybrid options where you can mix and match datacenter and virtual datacenter technologies.

Virtual datacenters represent the last model which represents fully virtualized datacenter resources delivered on-demand.  They are much like cloudcenters, but with much less control and options as the physical infrastructure is typically completely hidden.  The upside is that the level of complexity is significantly reduced.

Digging into each of these in detail will really highlight the differences.  I’m will dive into each below, but first I need to talk about the great unmentioned scaling problem.

Workloads.

Understanding Workloads
VMware’s documentation is littered with references to understanding server workloads because it’s such a critical issue for virtualization, yet it’s barely mentioned in the blogosphere or twitterverse when talking about cloud computing.  Simply put, server workloads define how a computing or storage resource is used.

Some workloads can be virtualized and some cannot.

Determining whether a given workload can be virtualized or not is actually relatively trivial.  Your workload is the application or set of applications that you need to run on a single server.  If this workload can use the total resources of a single physical server then it should not be virtualized.  When the workload results in a fraction of resource usage then you should move that workload into a virtual environment.  You must measure workload by many factors including CPU, RAM, and disk.

The most obvious example would be a large database.  Take any large database on a typical physical server today that has 8 cores, 32GB RAM, and a local RAID array.  If at peak usage that physical server is under heavy load you know that moving it to a virtual environment where it would have significantly less resources makes no sense.

This shows how not all workloads should be virtualized as you would wind up starving your application of valuable resources.

Now let’s see why this matters for each of the infrastructure delivery models.

Traditional Datacenters
Everyone is familiar with traditional datacenters and I’ve covered the topic a bit in the past in terms of how they are different from clouds, what is delivered in a typical datacenter, and more.  For the purpose of this discussion I’m lumping “hosting”, “co-location”, “internal datacenters”, and all of the other DIY datacenter solutions into one.  I’m also asserting that they are part of IaaS even though not traditionally seen as an on-demand or cloud service.  I think it’s important to consider traditional datacenters as part of the IaaS eco-system.  They are still an option and many of the typical ‘hosting’ providers essentially package up datacenter capabilities (storage, networking, power, and cooling) into something you can buy over the Internet using only a credit card.  Understanding this option is important to contrasting and comparing with cloudcenters and virtual datacenters.

Whether you build out your own datacenter or rent co-location space from someone there is a tremendous amount of work that needs to be done unrelated to your core product delivery.  This includes everything from the mundane (labeling cables, racking/unracking servers, replacing hardware failures) to the more specialized (configuring Internet connectivity, VLAN topology, internal network routing, virtualization strategies, etc).  For most businesses this work is a complete sideline to core features.  Only the largest web application providers (Google, Microsoft, Amazon, etc.) or datacenter-focused businesses (e.g. hosting providers) can really afford the time and resources to spend building and managing their own datacenters.

Datacenters are where the rubber meets the road in Internet and web application delivery.  They must be power-efficient, well maintained, and run professionally to be competitive.  A typical datacenter is built to provide a plethora of different services including physical servers, virtualization, storage, networking, sub-tenancy (aka co-location), server hosting, and more.  At the end of the day you can put anything in the datacenter that needs power and a network cable, regardless of what “it” actually is.

This means datacenters are ideal for customers who have unique or complex workloads that require them to really own and customize their environments.  The sky is the limit in a traditional datacenter.  You just have to pay for that flexibility in terms of capital and operating expenses.

Cloudcenters
An emerging trend pioneered by GoGrid and also pre-announced by RackSpace Cloud is the “cloudcenter“, a datacenter-in-the-sky.  Cloudcenters are an intermediate step between no virtualization and full virtualization.  They provide some of each, allowing maximum flexibility.  Like datacenters, cloudcenters will provide infrastructure services that encompass both the physical and the virtual.  A cloudcenter can allow you to run complex and mixed workloads.  Some cloudcenters even provide a co-location option for those who have very specific needs.

Unlike datacenters, cloudcenters typically deliver their services in an on-demand fashion allowing customers to avoid most capital expenses (CapEx).  There is also a direct impact to OpEx as well given that cloudcenters essentially package power, cooling, space, networking, and storage as services.  This obviates the need to hire the specialists to manage those parts of your infrastructure.

Recently, I wrote a whitepaper for GoGrid that talks about scalability issues with web applications that has a decided bias towards cloudcenters.  I invite you to read it as it highlights further how they are different from the other options.

Virtual Datacenters
The virtual datacenter was pioneered by 3tera, an early leader in providing software that created fully virtualized datacenter environments.  More recently, Sun Microsystems announced foray into cloud computing uses the virtual datacenter model as well and is likely to further popularize the approach.  While fully virtualized datacenters provide much less control and can only support limited sets of workloads they are the easiest to approach and consume for those who want to minimize both CapEx and OpEx.

Typical virtual datacenters provide servers, storage, and possibly some network control, but because virtualization technology is used at all layers they are currently unable to handle heavier workloads.  It is likely that as virtualization matures and Moore’s Law continues they will become the de facto standard.  Unfortunately, one major issue is that virtualization still needs more and better ways to guarantee service-level-agreements (SLAs) for CPU, RAM, and disk in multi-tenant environments.  The technology is still far from where it needs to be, although some (e.g. VMware) clearly have some capability in this regard.

Virtual datacenters still require at least a server specialist, but are the final step before that next layer of Cloud Computing called “platforms” or “Platform-as-a-Service” where even servers are fully abstracted from the customer.

Visualizing IaaS Options

The following diagram visualizes how the three IaaS delivery models work within the famous Cloud Pyramid:

Summary
So whether it’s a traditional datacenter, a cloudcenter, or a virtual datacenter that serves your needs, the important thing (as in all of IT) is to use the right tool for the job.  Determine your workload requirements and find the right solution for your needs.  When in doubt, I recommend the cloudcenter as the best interim solution.  Admittedly I’m biased, but after spending a lot of time running my own startup before joining GoGrid and looking at these issues in detail I’m convinced that it is ill-advised to spend significant resources on ‘horizontal scaling’ of database workloads until absolutely necessary.  As always, YMMV.

Hope this helps when trying to understand your cloud computing infrastructure options.