Cloud Services Accelerate Application Deployment

As cloud computing continues to mature and user experience grows, demand continues to increase for more sophisticated tools to further speed application deployment and reduce IT staff time. Vendors are addressing this growing need with deployment tools.

By David Jacobs

April 20, 2011

Rapid application development and deployment are always the desired goal for the IT staff, but it is more critical than ever in today's highly competitive and rapidly changing global business environment. The sooner an application is up and running, the sooner it can begin delivering new capabilities to customers or employees.

In the past, deployment was slowed by the significant time and effort required to install and configure the supporting environment. Both developers and system operations staff were required to invest time better spent on developing applications for business needs.

Cloud computing, both public and private, is proving a boon to IT staff and software developers. It speeds up deployment by reducing the time and effort needed to prepare the application environment.

Early cloud users benefited by saving the expense and time needed to determine whether existing in-house hardware configurations could support the new application and if not, to buy, install and configure additional hardware and system software.

As cloud computing continues to mature and user experience grows, demand continues to increase for more sophisticated tools to further speed application deployment and reduce IT staff time. Vendors are addressing this growing need with deployment tools.

These tools incorporate three types of cloud deployment patterns: virtual image, topology and workload cloud-deployment patterns. They offer users the flexibility to now choose the appropriate pre-configured environment for each application from among a variety of products and methodologies.

Virtual Image Cloud-Deployment Pattern

Early adopters of cloud services continued to deploy applications much as they had in the past. Each deployment required support staff to create a virtual image containing an operating system and all the other software needed to support the application. Still, they saved time by taking advantage of the immediate availability of cloud resources.

The first step these early users often took was to create one or more standard virtual images. Each image could be used repeatedly, so users saved time by eliminating the need to install components and configure an image for each application.

Vendors came to recognize the advantages of standard images and that common types of applications often require the same components. To address customer needs, vendors began offering their customers images containing a full set of components needed for these common application types.

Web-based applications are an example of a widely used application type and therefore an excellent choice for vendor-provided virtual images. Web-based applications are particularly well suited to the cloud. The virtually unlimited resources available enable the application to acquire and shed processor and storage resources as load varies over the course of a day.

Addressing these Web-based application requirements is the IBM WebSphere Application Server Hypervisor Edition virtual image, which incorporates the operating system, the Websphere Application Server and an HTTP server.

Even though Web-based applications share many of the same required components, there are differences. To address these differences, the IBM solution includes an additional component. The customization/connection function consists of controls that enable users to tune the image to meet the differing needs of specific applications.

To make use of the product, users activate the virtual image, passing in instance configuration data that is applied by the customization/connection function. Once the virtual machine is running, users simply log in, install their application, which then begins to execute. The image encapsulates the installation and much of the configuration of the operating system and middleware software. Application deployment is accelerated since there is no need to select and then install and configure compatible components.

Topology Cloud-Deployment Pattern

The virtual image cloud-deployment pattern eliminated a great deal of the time and effort required to deploy an application, but is limited to applications requiring only a single virtual image. While a single node is sufficient for some applications, many require multiple nodes, each with its own image.

Topology deployment patterns support applications that require multiple nodes. As configurations spread across multiple nodes, complexity increases, together with the need for additional time to assemble, configure and test the set of images, their respective interactions, and finally to deploy them. Vendor-supplied topology pattern products eliminate much of this complexity.

IBM's IBM Workload Deployer V3.0 supports both topology and workload deployment patterns. The topology option manages the application environment and provides security across a pool of virtualized servers, provides tools to ease virtual image customization, automates maintenance and ensures correctness and consistency for updates to virtual images.

Workload Cloud Deployment Pattern

Workload cloud-deployment patterns shift attention away from specific components and the way in which these components are distributed across nodes. Workload patterns move the focus to the application itself. Rather than select a set of required nodes, developers and support staff select a workload deployment package to suit the type of application.

For a batch application, IT staffers select a workload deployment package appropriate for batch; for a Web-based application, a Web-based package. In each case, the staff does not need to understand the application infrastructure that makes up the package, nor is it necessary to understand the integration or optimization techniques applied to that infrastructure.

Staffers do need to specify a set of application policies that control various aspects of operation. For example, the cloud is well adapted to applications that require varying amounts of resource during certain phases of operations or over the course of a day, but applications vary in terms of critical resource.

For some applications, increased load requires additional CPU capacity. For others, Web server or database server capacity is critical. Users create a policy indicating when or at what level of load resources are to be added and under what circumstances resources should be released. In addition to specifying how resources should scale, security policies must also be specified to define who can access the application and what actions each individual may take.

IBM Workload Deployer Pattern for Web Applications V1.0 is the first release of a series that will include workload patterns for a variety of application types. It operates with IBM Workload Deployer V3.0, and reduces staff effort and speeds deployment by:

Simplifying ongoing management by combining controls in a unified operator interface.
Reducing the possibility of error since the unified interface makes it impossible to make conflicting configuration changes in different components.
Aiding in reproducing an environment by maintaining policies, configuration settings and best practices in a coded form rather than requiring setting individual configuration settings for each new use of the environment.
Simplifies the use of a database by requiring little database configuration knowledge. Users simply load an SQL file that describes the database and tables their application uses.

In each case, beginning with the virtual image pattern, through the topology pattern and finally to workload, a greater proportion of the effort is shifted away from user staff. Some flexibility for customization and control is traded at each increasingly comprehensive level of standardization. Users gain at each level via lower total cost of ownership due to reduced time and effort deploying applications.

While each deployment pattern provides an immediate savings on the effort required for application deployment, those who adopt the workload pattern will reap an additional, apparent benefit: reduced staff training time.

Here's why: developers and system staff typically invest considerable time learning the characteristics of each new operating system release. The same level of investment is also required to learn the details on each of the other components. With the workload deployment pattern, the staff does not need to know the details of all the underlying components. Training time is greatly reduced.

Choose the Appropriate Deployment Pattern

There is no single deployment pattern that will meet every application development need. For a relatively simple application requiring just an operating system and a database, the virtual image pattern is appropriate. For more complex applications, the topology or workload patterns are more appropriate.

The virtual image deployment pattern and the topology pattern have proven to speed up application deployment and free IT staff from having to install, configure and test an array of components. However, a running application still remains the goal and selection of the appropriate virtual image or topology pattern represents just a means to that end. Among the issues associated with both patterns:

System staff and application developers must focus on the set of included components.
Staffers need a detailed understanding of the set of components needed to support the application.
Expertise is required to evaluate and select vendor product offerings.

In some cases, users may want more customization capability than either a topology or workload pattern will give them. In these cases, they will not be able to take advantage of vendor-provided images, but these cases are unusual. In the great majority of cases, users can take advantage of the flexibility provided by vendor supplied images and tools.

In any case, many applications will be able to take advantage of topology and workload deployment patterns, particularly Web-based application types. Workload deployment pattern products for other widely used application types will soon be available. As these products become more widely available and staff familiarity increases, enterprises will benefit by automating internal operations. The business and IT payoff: accelerated application deployment, greater efficiency, reduced IT infrastructure costs faster time to market for a greater competitive advantage.

Additional Resources:

Demo of IBM Workload Deployer

Blog by Dustin Amrhein

IBM Business Agility solutions

Cloud mini main tent at the IBM Impact 2011 conference

Join the IBM developerWorks Cloud Computing Group to connect and collaborate with technical subject matter experts on developing applications in the cloud

David B. Jacobs of The Jacobs Group has more than 20 years of computer and networking industry experience. He has managed leading-edge software development projects and consulted to Fortune 50.