The traditional focus of service virtualization has been modeling the technical requirements and dependencies between applications and their components for more efficient software development and management. Recently, Adaptivity has launched a software modeling methodology and tool called Blueprint4Cloud (BP4C) that combines business and technical information to help guide the strategic IT decision making process.
Cloud initiatives are typically focused on technical capability. Organizations will just look at what an application is doing technically, but they completely ignore what the business is asking for, said Jason Noel, vice-president and cloud practice lead at Adaptivity. “IT has never done a good job of understanding what the business is trying to tell them. A lot of times an app is moved to the cloud, without consideration for the other apps it is dependent on.”
Overcoming subjectivity with objective data and models
One of the biggest problems with manually modeling the fit between business software and technical requirements is that people are subjective, said Noel. “The biggest insight is that companies that think they know how their apps are deployed based on estimation are frequently inaccurate compared against a quantitative analysis of existing systems. Even when a customer thinks they know how their apps are deployed and the dependencies between them, we have never found someone to be entirely accurate. The best IT shops are at most about 80 percent right.”
Understanding dependencies has become more complex with the rise of SOA, as organizations have many smaller apps with a greater number of dependencies between them. The BP4C model helps organizations figure out how much applications are being used and their level of criticality to business processes. For example, if an organization has four similar applications being used by four separate groups, in many cases they can be consolidated into one. But this one app needs to fully meet the minimum business requirements of all four groups to facilitate a useful consolidation.
Historically, IT transformation takes months in going from design and deployment to a few days in the cloud. But the months that went into planning have not been sped up, argues Noel. The problem is that poor cloud implementation can create costly problems when the business people implementing them don’t understand that moving to a low-cost platform like Amazon might increase the latency of mission-critical process like customer service and end up costing the organization more in lost human productivity.
From Grid to Cloud
Adaptivity was founded in 2005 by several executives at Wachovia bank who had launched a grid environment, which was a precursor to modern cloud platforms. At the time, it was considered highly innovative with several thousand nodes. But the big problem was that after the grid was built, they did not know which apps they could economically move to the new platform. On the surface, many applications looked like they would be easy to move to the grid, but further analysis revealed that they were dependent on other applications that were not as easy to move.
Noel said they had to develop a methodology to look at a portfolio of apps to see if they were a good fit. This involved looking at thousands of interlocking sets of services in a systematic way. The team realized that an expert system approach could guide management in making better assessments of the potential and constraints of transitioning apps to the grid.
Over the past 18 months, Adaptivity has refined this methodology into a platform for guiding decisions for migrating enterprise apps to the cloud. Noel calls this practice a design science for modeling an enterprise’s software ecosystem.
This design science involves modeling numerical processing patterns, content collaboration patterns, and RIA patterns. The BP4C tool helps to identify the types of patterns being used by the organization and the components of these that matter to business needs. The design science underneath BP4C identifies the business importance of different applications and specifically what cloud platforms will be the best fit for a particular application. It reduces the up-front manual strategy and road map work by 60 to 80 percent, said Noel.
Noel explained, “It is not just about how much capacity you need, but how much storage, and the maximum acceptable latency. If I have a storage bottleneck, adding more low-cost storage will not solve the problem.”
The BP4C approach automatically generates a model illustrating the relationship between business applications, their resource requirements and the ways they address specific business requirements. Today, many large IT consultancies have developed methodologies for guiding these decisions manually. In these cases, teams of experts will guide an organization through a prescriptive methodology that involves interviews of different types of technical and business operations staff. But this process is resource-intensive.
Noel said, “A lot of people hire consultants, and an army of experts from IBM, Accenture or Ernst & Young will churn through the applications to understand how they are deployed and what the requirements are. It is all manually intensive. Then they will go in the back room and spend weeks or months to get thorough actionable recommendation for you, and then create a strategy for moving apps to the cloud.”
Speeding up the application modeling process
BP4C speeds up the enterprise application modeling process in two ways. First, it uses a variety of existing IT asset auditing tools to assess which applications are currently being used by the organization and the resources they are dependent on. Automated Dependency Discovery Management (ADDM) tools from vendors such as IBM, EMC, CA, and HP gather resource information about existing applications. However, it can be challenging to make sense of the flood of low-level information.
While these tools are good at capturing asset information, they don’t natively include ways of using this information for creating models that are useful for guiding the IT asset management process. “These tools are great, but they generate gigabytes of information,” Noel explained. “Instead of taking this information and doing data munching and spending weeks or months going through it, our platform will give a cloud fit score for each kind of information and it is all backed up by facts, or the requirements that business and technical staff have provided.”
The other component is a tool for Web-based question-and-answer style capture of the needs and requirements from different types of staff within the organization. This makes it easier to automatically capture the business needs in a way that can be precisely correlated with specific technical requirements.
There are two types of questions. Questions about the business look at the relative importance of CAPEX reduction, OPEX reduction, risk reduction and efficiency. Technical questions consider the workload and technical requirements of the applications.
The information from the automated enterprise analysis and the staff interviews are integrated into a systematic model that can be used to generate a cloud suitability dashboard. This can be used to calculate a cloud fit score and can help the enterprise to clarify what is pertinent to them. This helps the organization to get a better idea of the best candidates for the cloud.
On the output, BP4C can generate a full bill of materials estimating the number of virtual machine instances, storage requirements, and networking requirements to satisfy business needs in the cloud. Noel said they are also working on a set of tools for precisely calculating the precise cost of running these apps in the cloud. Adaptivity has implemented a preliminary cost analysis tool, but they are still refining the algorithms. Noel said, “When it comes to pricing and you are telling the organization you are going to save a certain amount of money, you’d better be right.”
In the long run, better business software modeling tools like BP4C promise to complement the more technically oriented approach of service virtualization tools. Organizations will be able to identify business dependencies more quickly with business modeling, and then to better craft software modules to meet these within the enterprise or the cloud more quickly and reliably.