Scientific visualization has matured over the last two decades with the accumulation of knowledge and expertise from various field of research. High performance visualization is therefore built on top of various related underlying methods and techniques from different fields. Usually, the HPC (High Performance Computing) users have more specific needs than general ones, and besides the formulation of new visualization methodologies and approaches, it becomes crucial to actually implement and deliver these visualization capabilities to the end users. This development process is usually a long-term activity thus it requires a long-term software development and maintenance process. Therefore, the system design is required to cope with the dynamic changes in the computer architecture of the HPC systems over the years. A flexible visualization framework which can cope with these architectural changing, can interact with different hardware devices, and can easily be aggregated with other visualization methodologies becomes highly necessary. In addition, it is desired to be capable of adapting and coordinating with different post-processing requirements related to the changing in the use of HPC resources for better efficiency. We will describe the top three issues related to the design and development of this kind of visualization framework to deliver ubiquitous and sustainable High Performance Visualization capabilities.