data partitioning : R-tree
denser data. dont scale with senser data
mesh that simulate brains neurons
R-trees minimum bounding rectangle
sequence read archive
data driven science: data drives where scions goes,
wasted data -> worst ROI! delete data cause you don’t have room for it
find interesting data, move quickly through data, how to enable scientific discovery
spatial queries on unstructured mesh
tetra hydra how earthquake propagates
store info on nodes
point: which tetra hydra this point belongs to
index data that might be only intereting dont index everything.
on demand tacking moving data
data, knowledge, patterns, rules, unifying models.
run queries to create datasbe--- dont create datasbe to run queries.
CHANGE THE WAY WE LOOK AT DATA SYSTEMS
IN0SITU QUERY ENGINE
we create a datasbe and ask neuroscientists earthquake sceintists come and use it. let them lead and we adapt make it fast.
bring mutilple bridges to scientists.
scidb operational there with nodb: no data.
why is scidb hard? why is it not invisible.
deep natural language processing
What MapReduce can't do Limitations of Mapreduce – where not to use Mapreduce