As we move in to a cloudier and cloudier IT landscape, defining the workload is going to be more and more important. It’s critical not only for reducing cost (what services can this workload do without? How much capacity?), but also for increasing performance and agility.
A workload is a tricky thing to define and there are a number of strategies for how to do it. Some will have you look at the servers that a workload is running on and the capacity and utilization of each one. Some will have you interview application owners about requirements. Others would have you snap your workload into a template of some sort (e.g. it’s a Ruby Server). All of these have some validity; they’re great ways to determine capacity, non-functional or functional requirements, but they don t let you see the whole picture. They don’t codify the mass-customizable world of IT that we live in today. Now is the time to build systems that are not only efficient and low cost, but also meet high-performance and fit-for-purpose needs for particular workloads.
With this in mind, Adaptivity’s Design Studio allows you to understand your workloads across 51 different characteristics and use this knowledge to make decisions about what type of cloud environment to place each of them on. We do this not only by understanding the application (are we talking about your research tools or your content management system?) but also the parts of the business that rely on the workload (is it leveraged by your 24/7, customer facing stock brokers or your 9-4 tellers?). All of that knowledge can be combined and made sense of using Design Studio.
The business knowledge is gained by selecting the business value chain (BVC) activities that the workload supports. (Click here for a Use Case Example on Building the BVC). In our Design Studio Classic Tool you can see all of the activities in your organization and place check marks next to the applicable ones. For example, a workload responsible for notifying customers about new products would support a business activity around customer relationship management and one around inventory.
The information about the type of workload is gained by matching it to a pattern. Adaptivity has defined over 50 different workload patterns. In the case of our new product alerting system, we would likely select a Content Distributor Pattern.
Once we’ve selected both the pattern(s) and the part(s) of the business supported, we must balance the conflicts between the disparate sources of information (the application and business requirements). For example, inventory management workloads usually have very fine-grained workload units (often just database records) while content distribution patterns usually have much coarser grained workload units (often unstructured documents or pictures). The Design Studio tool addresses these differences in three distinct ways (all using the “Tune” dialog window that can be accessed by pressing the “Tune” window).
- Taking the “average” quality requirement for every quality. This will go through each quality and assign a value that is indicative of the entire set of business activities and patterns being considered. For example, it would suggest “medium” level workload units to accommodate for the fact that the Content Distributor Pattern generally uses coarse-grained workload units while the inventory management activity often uses fine-grained ones.
- Taking the “maximum” quality requirement for every quality. This will go through each quality and select the value that will be most difficult to provide infrastructure for. For example, it would select coarse-grained because it is often more difficult to deal with than finer-grained records. This is usually only done for very mission critical workloads.
- The best result is always to go through each quality and carefully consider your answers and selections. In this case, when we consider a workload that sends information to customers, we are probably considering one that uses pictures and unstructured data. So we should probably select the coarse-grained option.
Once we’ve done this we have an accurate picture of the workload and are in a much better place to select either a cloud provider or an internal infrastructure option to facilitate workload execution. Design Studio can accomplish this too by leveraging Adaptivity’s proprietary calculation data.