Demystifying the cloud

Demystifying the cloud: A practical roadmap to deploying cloud computing

Cloud computing is often billed as the next big step in organizations’ perpetual drive to gain greater efficiency and agility from their IT resources. In fact, many IT professionals feel it’s inevitable that some part of nearly every organization’s network will be moved to the cloud. But at the same time, confusion persists: What exactly is cloud computing and is it the right move to make now?

Cloud computing allows organizations to access software, computing power and application development tools from third parties over the Internet. Its greatest benefit is the agility that comes from being able to add or subtract massive computing resources at a few minutes’ notice. While the benefits are significant, the wisdom of outsourcing core business processes to the cloud today depends heavily on your organization’s business model and needs.

This white paper will show you the essential components of cloud computing, how it’s being used to great effect today and how to create a roadmap for a successful implementation. You’ll discover:

• The tremendous benefits of cloud computing
• How to create a step-by-step cloud implementation roadmap for your organization
• What to look for when choosing a cloud service provider
• How cloud computing has helped organizations such as The Washington Post and others

1.0 Cloud computing defined

Cloud computing is a catchall term that means leveraging the Internet ‘cloud’ for business computing services. With cloud, computing services are supplied over the Internet from scalable data centers made up of thousands of virtualized processors. Workloads are distributed across multiple servers, which can potentially be located anywhere in the world. By outsourcing their computing, organizations transfer the burden of software and hardware financing and management while simultaneously increasing their ability to scale to suit changing business conditions.

While most cloud computing solutions make use of geographically disparate data centers, managed or hosted services provide a kind of private cloud computing whereby customized applications are hosted in a specific data centre, and can be run on dedicated servers. In this case, outsourced services are accessed over the Internet, but are not shared with other clients–an arrangement important to organizations with high data location and security priorities.

While the definition of cloud is a hotly debated topic, most experts generally agree that in order for a service to qualify as cloud computing, it must:

• Change capital expenses into operating expenses
• Remove the burden of hardware management
• Be available, in unlimited quantity, whenever you want it
• Follow a pay-per-use model, with no contractual obligation

All cloud services share these traits, but they differ greatly in their offering. There are three main kinds of cloud services:

• SaaS – Software as a service consists of applications such as billing and CRM software, provisioned over the Internet
• IaaS – Infrastructure as a service provisions infrastructure such as storage and computing power
• PaaS – Platform as a service furnishes complete application development environments, eliminating the need to build interfaces and administer platform growth and other factors

1.1 The benefits

The strongest benefits of cloud computing include:

• Agility – Adding or subtracting any number of services in a matter of hours
• Scalability – On-demand provisioning with no requirement to engineer for peak loads
• Cost – Replace capital expenditures with operating expenditures on a pay-per-use basis. Decrease or eliminate other costs as well: real estate, HVAC, electricity and maintenance, security and more
• Location and device independence – Access systems over the web regardless of location

IaaS–what most people think of as cloud computing–is where most organizations are currently finding the greatest benefits. Suppose you make software for creating music videos, and subscriptions skyrocket overnight, requiring a sharp increase in computing power. With cloud computing, servers can be automatically and sequentially added as needed, meeting demand without interruption.

1.2 Specific requirements

While cloud services are improving in many ways, some organizations still find that their specific requirements do not mesh with what cloud computing currently has to offer. In most cases, applications need to be designed to specifically leverage the features of the cloud. For example, traditional ‘boxed’ software business applications are tightly coupled to hardware, and as such they are not suitable for use in the cloud. In addition, some cloud providers cannot guarantee where data will reside, or that it will be adequately secured.

Another thing to keep in mind while considering outsourcing to the cloud is that speed of service depends not only on the speed of servers, but also on network connections and Internet routing. For this reason, hosting proximity and other location-specific factors are still important - particularly for applications with video or voice traffic. Finally, if the volume of an organization’s outsourced computing requirements do not fluctuate, cloud computing solutions may not make economic sense. Because IaaS and PaaS provide computing power in any amount on demand, cloud rates for infrastructure and platforms are generally greater than those available on a long-term contractual basis.

2.0 The cloud computing landscape

Cloud computing is a relatively new way of expressing a concept that has been around for some time. As consumers, we have been using SaaS applications like Hotmail and other Web-based email since the 1990s. Many organizations have also been using SaaS in some capacity since the early 2000s.

The first cloud offerings to take hold in the business world were SaaS applications such as salesforce.com, a hosted customer relationship management platform, and Zoho’s suite of business productivity applications. The SaaS category continues to grow quickly: new tools like Microsoft’s Business Productivity Online Standard Suite (BPOSS) show great promise in eventually replacing robust boxed business suite solutions, simplifying the role of IT departments by eliminating the need for in-house software installation, patching and upgrades.

In many ways, the features of SaaS cloud services aren’t new at all, but are merely a rebranding of the ‘application service provider’ model of the late 1990s. However, the key feature being leveraged today is that of scale, thanks to multi-tenancy – many users and companies sharing one massive instance of the same piece of software.

The next category of cloud services to emerge–and what most people associate with cloud computing–was Infrastructure as a Service: computing and storage on demand. Two of the best-known IaaS cloud computing providers include Amazon.com and RackForce, both of which can provision petabytes of computing power in minutes. The business community first became acquainted with IaaS when Amazon–at the time an online retailer that conducted a disproportionate amount of its yearly business in the span of a single month–decided to see if it could rent out its massive server capacity that sat idle much of the year. The answer, as we know, was a resounding yes.

Platform as a service, the final category, is just beginning to emerge. Many believe that PaaS is the ultimate evolution of cloud computing. Middleware, database and web server connections and configurations are fully managed, leaving you with nothing more to do than to code an application on the ready-made platform and launch it from your virtual, cloud-based machine. PaaS is very promising from an efficiency point of view, and likely to quickly gain traction once a standard model is established for the service. One of the best known examples of PaaS is Microsoft’s Azure®, a set of cloud computing services ready to be used either independently, or in conjunction with in-house software applications.

2.1 Cloud computing today

A broad spectrum of businesses already uses CRM software, online timesheets and other kinds of software as a service. But while SaaS can be useful to many businesses in a variety of situations, IaaS and PaaS applications are more highly specialized. With IaaS in particular, cloud computing currently makes best sense when:

• You have short-term needs for capacity, such as seasonal demand or resource-intensive marketing campaigns
• You are experiencing unexpected demand or traffic
• More firepower is needed for short-term testing and development
• Money or existing capacity are tight
• Provisioning computing needs as operating expenses rather than capital expenses makes sense

Using the cloud for added firepower works well as a short-term strategy and is vastly cheaper than alternatives. But if you are going to require IaaS resources in the longer term–for more than 12 months, for example–in-house resources or traditional data centre outsourcing can prove less expensive.

Diagram: Cloud computing costs

2.2 How cloud computing is changing

Cloud offerings are growing richer as cloud computing gains traction. More players are working quickly to make the cloud a more hospitable environment, resulting in some positive changes. The most significant thing to happen in the near term with cloud will be an expansion of features. Not long ago, IaaS was limited to processing power, memory and disk space. Interfaces were extremely limited in terms of configurable features–you could not even control firewall rules. Now you can manage your firewall rules and drag-and-drop a lot of the features you want, such as server templates, scripts and input parameters. Some providers even allow you to automatically distribute your content via their own content delivery network (CDN).

More customization

Another area of growth is application customization. SaaS in particular works best with applications that are widely used and broadly homogeneous: CRM, HR and email software are the most common examples of these. But systems that are more highly tailored to the needs of individual organizations, like billing, project management and ERP, are increasingly being serviced by the cloud.

Customized applications are proliferating especially quickly within SaaS vendor stores. Salesforce.com and others have launched business-to-business application stores where hundreds of applications are available for download and further customization. Many of these create workflows to help automate work processes and improve efficiency. At the moment, these applications only integrate with same-vendor software. When they graduate to the open data sphere and can work with any vendor’s applications, as they are just beginning to do, businesses are going to make more regular use of a greater variety of applications.

Cloud-ready apps

In the next 12 months, more cloud-ready applications will proliferate. While most software in business use today is tightly coupled to hardware and can’t be used in a cloud environment, more applications are using an Application Program Interface (API) so that they can be provisioned over the Internet. As more applications are brought online in this way IaaS, or renting servers, will make more sense for a broader spectrum of organizations.

Better security

The last major short-term development is the move among trusted providers to provide cloud services in a secure environment. That means that the same sort of security levels and guarantees now available in domestic data centers will be extended to the cloud.

2.3 The future of the cloud

Over the next couple of years organizations will increasingly orient themselves towards the cloud. What this will encourage in the longer term–say three to five years–is federation between cloud providers. Federation will allow applications and data to flow in much the same way as people consume electricity today. If your workload is going to spike, you can be assured of automatic capacity. Switching back and forth between trusted infrastructure providers may even become automatic, as conditions in the cloud change.

The rise of PaaS

Platform as a service will slowly gain strength. In three to four years’ time, we will see PaaS taking configuration out of the picture entirely. Instead of needing to build and configure an entire platform of servers, middleware, database, etc. on which an application is to be developed, developers will be able to concentrate on building applications in a ready-made environment without needing to take care of–or even know–what is going on underneath the service.

This bifurcation of technical configuration and application-building–the end goal of PaaS–is particularly important from the point of view of resourcing and scalability. With PaaS, there will be less need for people with system administration skills, and the process of scaling up computing resources will be handled as a service, in the manner specified by sales level agreements (SLAs). PaaS will also begin to offer complete lifecycle management for software development, including planning, design, building applications, deployment, testing and maintenance.

In the longer term, we will see a significant increase in new applications being designed on PaaS, particularly those that concern front-end customer interfaces such as social media, marketing or awareness campaigns where a customer can interact with an organization online. These applications frequently need to be refreshed, and we will increasingly see them being placed on cloud-based platforms.

Choose your security

As cloud computing matures, more providers will obtain security certifications. In fact, certification will become more harmonized, with standards such as Payment Card International (PCI) and Statement on Auditing Standards No. 70 (SAS70) becoming generally adopted. This will help people become more comfortable with outsourcing sensitive material to the cloud. In two or three years’ time you will be able to choose your level of security for every cloud application, priced accordingly.

3.0 A roadmap for cloud computing implementation

Cloud computing shows tremendous promise. In many ways, it is the next major step in the evolution of the Internet for business applications. As it is still emerging, the cloud landscape is fragmented and offerings differ greatly. Fortunately, there are clear best practices in making the move to cloud computing. Here are the seven steps to take when planning your organization’s entry into the cloud.

3.1 Consider your workload

Some kinds of work are better suited to cloud computing than others. Is latency going to be a big problem? Can you tolerate any downtime? What about security concerns? Cloud computing SLAs don’t typically carry the same guarantees as those you would get internally or with a traditional hosted facility. Because most cloud providers are competing on price, they generally don’t build systems engineered for high levels of uptime. And if latency is an issue for your organization, you will want servers close by, with a strong pipe between you and them. Unless data centres and cloud providers are PCI- or SAS70-certified–depending on your requirements–sensitive data should not be hosted in the cloud.

3.2 First, virtualize

Data centre virtualization is not a prerequisite for cloud computing. However, if you have not yet virtualized, you may find that it frees up server capacity that you didn’t think you had. Virtualization uncouples applications from underlying hardware, allowing you to optimize hardware for the workload.

Ten years ago, each server in an enterprise environment generally ran one application–witness mail servers, web servers, etc. Applications typically massively underutilized the hardware, making use of between 20% and 50% of available processing power. Once a server is virtualized, it can be moved from one piece of hardware to another: it’s literally a drag-and-drop procedure. This greatly reduces hardware, electricity, real estate and other costs and makes it easier to take advantage of cloud resources.

3.3 Do the math

Is cloud computing the most economical way to provide the computing power or software you need? As a general rule, if your needs are short-term and you don’t know how much usage you will require, cloud can make great financial sense. To work out the economics, compare:

• Capital versus operational expenditures
• Contract versus cloud rates
• The likelihood of requiring computing power on an ongoing basis
• The provisioning costs of renting cloud space versus an in-house solution, including capital expense, expertise, real estate, electricity, HVAC and other costs

Finally, carefully examine your SLA. Terms and conditions vary widely and upon examination they can sometimes reveal that a good deal is not such a deal after all.

3.4 Find out if applications are cloud-ready

Many applications have a one-to-one relationship with the hardware that hosts them. Such applications can usually be installed on a virtual server, but can’t be extended to more servers. Google’s Gmail is one of the first cloud-ready applications–one where the application controls the underlying equipment. If a server dies, a data centre goes down or workload decreases, Gmail automatically moves its workload over and shuts servers down as needed.

Take a look at the applications that you are planning to host in the cloud. Can you leverage APIs when designing your program so that the application can automatically grow or shrink, and call in resources as required? If not, cloud is not the answer. But if the benefits of cloud are just too strong to pass up, consider upgrading to a newer version of the application or migrating to one that is cloud-ready.

3.5 Compare providers

Does being hosted in Canada matter to your organization? Do you care if a cloud provider also offers secure, persistent storage, strong service and a host of technical features? No two providers are the same, so due diligence during the selection process is key. This might seem like common sense, but many companies were leveraging the Amazon cloud for credit card transactions before Amazon introduced a PCI-certified payment service.

Factors to consider when choosing a service provider:

• Latency – If speed matters, you will need to look at the latency between your customers and the data they need to access. Are processing speed, possibly QoS and connection guarantees, on offer?
• Security – How sensitive is your data? Some providers offer PCI and SAS70 certification. How data is accessed also affects security and latency–over the public Internet, a VPN connection or other certified, private circuit. Will you be able to monitor and audit where your data is located? Redundancy will also affect data security
• Interface and service – It’s much easier to deal with a cloud provider that has good customer support, a great portal, billing that’s easy to understand and superior technical and service features. What is the quality and availability of reporting? Can your users call their support staff, or is cloud provider support for secondary use only
• Total offering – Are you looking for a one-stop shop? In addition to cloud computing, some providers also offer CDN, storage and other capabilities, making it easy to consolidate services with one provider
• Contract terms – The SLA that you sign should take all of these factors into account. Keep in mind that contract guarantees need to be adequately backed up. For example, the penalty a provider pays for downtime should be commensurate with the effect of the interruption on your end. Check past performance as well–obtain records of outages on a historical basis and drill down as necessary in other areas, such as customer support

3.6 Proof of concept

If cloud computing is new to you, it’s best to run a pilot program. Hard costs will be few; a pilot will demand more time than money. The key is to put meaningful but non-critical work into the cloud under real load, testing pre-established processes and working through mock scenarios similar to those that you would encounter in a permanent cloud rollout. For example, in a CRM system scenario, manipulate the data and generate sample reports.

Make sure to test everything that will impact the quality of the entire experience, whether it’s customer service or load speed. For example, consider testing:

• How quickly new servers can be added, transactions take to process and invoices are issued
• Adding users–will this affect application performance?
• Features and services, in terms of configuration–firewalls, network management and more
• Quality and availability of reporting–can your users call their support staff, or is cloud provider support for secondary use only?
• Monitoring–can you use internal data monitoring tools?
• Setting thresholds for scale–can you prevent the pilot from expanding to thousands of servers?

3.7 Migration plan

If proof of concept results are positive, it’s time to migrate data over for use in the cloud and to train staff. Introducing users to a new technology used to be a bigger issue. Today, many employees will have grown up with more computer applications in their personal lives than they have in the workplace. The bigger challenge lies in migrating data. For this reason, it’s often wise to engage a partner that specializes in data migration.

4.0 The cloud in action

There are thousands of examples of cloud computing in action. The following are a few actual cases that highlight current best uses of the cloud to produce tremendous results.

4.1 The Washington Post: turning hundreds of hours’ work into just nine

From image to text

Peter Harkins is a senior engineer at The Washington Post who found out that Hillary Clinton’s official White House schedule was soon to be released from the National Archives. The information was due to arrive the next day: 17,481 pages of non-searchable PDF document detailing Hillary Clinton’s daily activities as First Lady during President Bill Clinton’s two terms in office. Harkins calculated that it would take hundreds of hours to pore through the document’s low-quality PDF files–too long for a news scoop. He needed to find a way to convert the images into usable, searchable text and deliver them to the newsroom the same day.

Harnessing the cloud

Harkins first tested various PDF and optical character recognition tools to convert the images into machine-readable text. He soon found that it was going to take about 30 minutes per page. Working against time, Harkins moved the project to the cloud, in this case Amazon’s EC2. He launched 200 server instances to process the images to his own custom specifications.

Results on time

With a processing speed of approximately 60 seconds per page, the project was completed within nine hours and sent to writers, who began searching against the data. 26 hours later, Harkins and his team launched a polished web interface to make their searchable database available to the public. Harkins concludes: “EC2 made it possible for this project to happen at the speed of breaking news. I used 1,407 hours of virtual machine time for a final expense of $144.62. We consider it a successful proof of concept.”

4.2 Animoto: overnight success

Steeply spiking demand

Animoto Productions’ popular web application uses editing software to automatically generate professionally produced videos based on user-selected images, videos and music. Animoto found overnight success when it launched its service as a Facebook plug-in, driving incredible demand and rapidly spiking the load on back-end resources.

Automatic provisioning

To keep up with demand, hundreds of Amazon.com EC2 servers were automatically and sequentially added to the 50 servers that Animoto was already using, resulting in a peak of approximately 3,500 servers in action. Of course, servers were added and subtracted as demand grew and dropped.

All users served

All user requests continue to be served, with no intervention necessary on the part of the Animoto team.

5.0 Reach for the sky

As cloud computing gathers steam, more providers are improving services, infrastructure and platforms, making the cloud a more hospitable atmosphere for a growing number and a greater variety of organizations. Thanks to the increased agility and scalability that come along with it, cloud computing can represent a tremendous strategic opportunity and deliver serious cost savings. But because some applications are not cloud ready and provider security and performance vary widely, cloud does not make sense in every situation.

Software as a service is where the broadest range of organizations can take advantage of robust third party applications, so it comes as no surprise that it’s currently the greatest area of growth. Infrastructure as a service is also gaining popularity, as it gives the ability to add or subtract massive amounts of computing power within minutes. But the greatest opportunity in cloud computing lies with platform as a service, truly in its infancy at present, because it enables developers to create and run applications without dealing with the logistics of building an interface, and without having to worry about scalability issues.

If your data must be treated securely, requires robust back-up and performance and downtimes are major issues, your choice of cloud computing provider is important. Not all providers can guarantee redundancy and security levels, or even where your data will reside, so it’s crucial to scrutinize SLAs when selecting a partner. As the cloud landscape matures, we will see more trusted providers offering cloud computing services, a broad range of security options, better connectivity guarantees and more detailed SLAs.

Talk to Bell

Bell offers a full range of cloud computing services and acts as a trusted advisor to medium and large-sized organizations across Canada. With Bell, your computing resources are directly connected to a private, highly secure and SAS70-certified MPLS network–eliminating many of the risks now associated with cloud computing.

If you would like to learn more about our cloud services, contact your Bell representative or click here to have a Bell representative contact you.

About the author

Strahan McCarten is Director of Product Management for Data centers and hosting services, including co-location and managed hosting with Bell. He has 10 years of IT management experience in both technical and sales roles, as well as 3 years in a consulting capacity. Strahan has written articles on cloud computing and spoken frequently at vendor and analyst conferences, and leads Bell’s Cloud Computing product teams.