By

Over the past decade, the pervasive use of object-oriented languages and the in-creasing complexity of problems solved by computer software have led to the prolifera-tion of large-scale framework-intensive applications. These applications are typically built by combining standard packages (e.g., J2EE application server frameworks), third-party layers for domain-specific functionality, and in-house solutions. While employing libraries and frameworks eases the development effort, it can be expen-sive to invoke the general APIs provided by these libraries, especially when they are used for simple tasks. As a result, many applications suffer from excessive memory footprint caused by chronic runtime bloat that significantly impacts scalability and performance. In addition, programmers are taught to pay more attention to abstrac-tions and models, and to leave the performance optimization to the runtime system. While a few redundant objects, calls, and field copies may seem insignificant, the problem quickly gets exacerbated due to nesting and layering. At some point, these inefficiencies accumulate, and compilers (e.g., the JIT in a JVM) can no longer elim