Migrating a Complex Industry to Cloud

Abstract

Many established industries face a dilemma about Cloud migration, as the ROI (Return on Investment) of such an effort is not clear. The EDA (Electronics Design Automation) industry’s software, tool methodologies, and flows and scripts for VLSI design have evolved from individual programs on mainframe computers, through collections of tools on engineering workstations, to complete suites of tools with associated methodologies on networks of computers. Design automation is one of the reasons why computer chips with upward of billions of transistors can be designed. One can assert that the server farms that form the back end of the Cloud would not have been around without the EDA industry. Thus, it is interesting to see whether Cloud Computing can, in turn, facilitate future growth of the EDA industry. Based upon a previous categorization of Cloud Computing workloads, this chapter maps the sub-tasks of an example silicon design flow to the types of workloads. The mapping of workloads is applicable to both Private Cloud Computing and Public Cloud Computing. This mapping can serve as an example for EDA companies and hardware design firms as they look to explore the Cloud for hardware design tasks. Our method can potentially open new doors and customer bases for enabling EDA growth. This chapter also provides examples of some early adopters, the issues they faced, and new emerging challenges, whether real or perceived. Additionally, some considerations are mentioned, such as licensing and delivery mechanisms that go beyond the mapping of tasks to workloads. The major contribution of this chapter is a proposed method for mapping EDA tools to Cloud Computing categories to facilitate the decision of which EDA tools are candidates for moving to the Cloud. Such a step is needed to migrate any established industry to adopt Cloud Computing.