Dave Durkee recently attended the Federal Cloud Computing Summit in Washington, in which a number of Cloud Computing vendors met with Vivek Kundra, the national CIO, and the CIOs of various government agencies.   We'll have a blog article about some of his impressions (he was very impressed with the government's desire to learn about and use Cloud Computing) at a later point.  The definition of Cloud Computing that the discussion centered on was surprisingly complete, broad, and useful, and I'm including it below.  

The source of the following working definition is the National Institute of Standards and Technology (NIST) Information Technology Laboratory (ITL). The final version is to be published in the upcoming NIST Special Publication on Cloud Computing and Security.

Definition of Cloud Computing:

Cloud computing is a pay-per-use model for enabling available, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, services) that can be rapidly provisioned and released with minimal management effort or service provider interaction. This cloud model promotes availability and is comprised of five key characteristics, three delivery models, and four deployment models.

Key Characteristics:

On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed, without requiring human interaction with each service's provider.

Ubiquitous network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs).

Location independent resource pooling. The provider's computing resources are pooled to serve all consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. The customer generally has no control or knowledge over the exact location of the provided resources. Examples of resources include storage, processing, memory, network bandwidth, and virtual machines.

Rapid elasticity. Capabilities can be rapidly and elastically provisioned to quickly scale up and rapidly released to quickly scale down. To the consumer, the capabilities available for rent often appear to be infinite and can be purchased in any quantity at any time.

Pay per use. Capabilities are charged using a metered, fee-for-service, or advertising based billing model to promote optimization of resource use. Examples are measuring the storage, bandwidth, and computing resources consumed and charging for the number of active user accounts per month. Clouds within an organization accrue cost between business units and may or may not use actual currency.

Note: Cloud software takes full advantage of the cloud paradigm by being service oriented with a focus on statelessness, low coupling, modularity, and semantic interoperability.

Delivery Models:

Cloud Software as a Service (SaaS). The capability provided to the consumer is to use the provider's applications running on a cloud infrastructure and accessible from various client devices through a thin client interface such as a Web browser (e.g., web-based email). The consumer does not manage or control the underlying cloud infrastructure, network, servers, operating systems, storage, or even individual application capabilities, with the possible exception of limited user-specific application configuration settings.

Cloud Platform as a Service (PaaS). The capability provided to the consumer is to deploy onto the cloud infrastructure consumer-created applications using programming languages and tools supported by the provider (e.g., java, python, .Net). The consumer does not manage or control the underlying cloud infrastructure, network, servers, operating systems, or storage, but the consumer has control over the deployed applications and possibly application hosting environment configurations.

Cloud Infrastructure as a Service (IaaS). The capability provided to the consumer is to rent processing, storage, networks, and other fundamental computing resources where the consumer is able to deploy and run arbitrary software, which can include operating systems and applications. The consumer does not manage or control the underlying cloud infrastructure but has control over operating systems, storage, deployed applications, and possibly select networking components (e.g., firewalls, load balancers).

Deployment Models:

Private cloud. The cloud infrastructure is owned or leased by a single organization and is operated solely for that organization.

Community cloud. The cloud infrastructure is shared by several organizations and supports a specific community that has shared concerns (e.g., mission, security requirements, policy, and compliance considerations).

Public cloud. The cloud infrastructure is owned by an organization selling cloud services to the general public or to a large industry group.

Hybrid cloud. The cloud infrastructure is a composition of two or more clouds (internal, community, or public) that remain unique entities but are bound together by standardized or proprietary technology that enables data and application portability (e.g., cloud bursting). 

Each deployment model instance has one of two types: internal or external. Internal clouds reside within an organizations network security perimeter and external clouds reside outside the same perimeter.

Note 1: Cloud computing is still an evolving paradigm. Its definitions, use cases, underlying technologies, issues, risks, and benefits will be refined in a spirited debate by the public and private sectors. These definitions, attributes, and characteristics will evolve and change over time.

I keep reading articles about how Cloud Computing is "high performance," especially from vendors and their partners who are eager to sell it.   Unfortunately, these discussions don't typically define what "high performance" is.  For example, today's excellent article on Cloud at Network World gave performance only a cursory discussion at the end.  Many people assume that today's microprocessors are so powerful that today's Clouds can handle any application, but my experience with enterprise applications is that overall performance depends on a lot more than just CPU speed or RAM size, and is very dependent on exactly what the demands of the application are (and therefore the actual design of the application as well.)

What I've seen is that discussion of cloud computing performance often avoids difficult topics that cloud vendors would rather not discuss because it might tarnish the current impression that Cloud is high performance and suitable to every need.  While Cloud deployment can benefit a majority of applications, it isn't suitable to all though I believe it will be eventually.  Let's take a look at some of the performance factors affecting the Cloud.

Latency (between the user and the Cloud or between elements in the Cloud) is often the only performance factor that is discussed, and it is important: we get customers from Amazon that talk of seconds of latency in some cases.  But as many in the industry are quick to point out, latency is dependent to a significant degree on the "last mile" of network leading to the consumer/viewer of a cloud service and not under control of the cloud provider at all.  I saw this clearly when I was working at NetSuite, as we struggled to provide good response times across the entire United States.  We were able to do so by carefully selecting our data center and bandwidth providers, and working with them.  Is this done by all Cloud vendors?  I don't know.   For enterprises considering replacing internal computing with Cloud and already having well-designed internal networks, latencies will almost always be less accessing internal resources than a public cloud because the network paths are finite in number and someone has been tasked to resolve the associated latencies.

However, there are other performance factors beyond latency.  To achieve that bottom-dollar price, most public clouds are built on commodity servers and networking equipment that may be adequate for a majority of applications, but are wholly incapable of addressing the needs of applications with very high I/O, memory, or CPU requirements.  For customers coming to ENKI from other clouds, we've seen I/O wait percentages of 80% or more of subscribed CPU time for some applications, particularly those with intense database loads or interconnected instances.  In other words, 80% of the cloud subscribers' paid resources went unused due to I/O delays!  Without knowing the architectural limitations of proprietary public cloud architectures such as Amazon, I surmise that these delays could result from congestion at the shared physical server's network interface, or from storage performance requirements that were not met by the cloud provider's storage offering - among other causes.  In general, as the speed of processors has outstripped that of storage and I/O, these problems have gotten worse.  But the use model of Cloud, which fully occupies the hardware through virtualization and shared usage, means that they can become critical.  Solving some of these problems so that Cloud users will be able to take advantage of today's faster processors will be expensive since they will require radically different hardware infrastructure than current cloud providers offer, which I expect will result in significantly higher prices for high performance Cloud Computing.

In the meantime, working with a vendor that can assist you in adapting your application to the cloud can reduce the impact of any potential performance bottlenecks by taking advantage of the underlying systems' strengths.

I don't think these limitations are a blot on Cloud Computing itself (which I'm of course excited about!), but rather a call to further discussion, development of higher-performance cloud infrastructure, and of course a frank discussion by cloud vendors of the limitations of their solutions.  This will avoid customer disappointment and frustration as well allowing cloud vendors to focus on their customer's needs. Avoiding this discussion concerns me that Cloud will be labeled as an over-hyped product.

Here at ENKI we've seen the need for a new hardware architecture that can provide radically higher performance to our customers, and we're in the final stages of qualifying it.  It will provide radically enhanced I/O bandwidth by speeding connections between our servers and to storage, as well as offering greatly enhanced storage speeds. This will allow our customers to deploy applications to our Cloud that formerly would have required a custom-designed datacenter and the associated costs and skills.