2012 Unidata NetCDF Workshop > The Future of netCDF
36.4 Speculations
Some timid(?) speculations about longer-term scientific data access infrastructure
-
I/O bottlenecks for HPC worsen
-
Use of massively parallel shared-nothing file systems spreads
-
Data generated too fast to store, filtering becomes priority
-
Multi-resolution wavelet representations get popular
-
Non-volatile memory technologies replace disks, change programming
-
Lack of organizational support leads to loss of Valuable Data
-
Format-independent conventions evolve too slowly
2012 Unidata NetCDF Workshop > The Future of netCDF