Hi
Mathias,
Great
question. You’re right, each position is a feature in a point
data source but you can also store the whole trajectory as a
linestring feature if you so choose. For a point feature, you
can have an attribute that threads point features together
into a trajectory and query using that attribute for the whole
trajectory. GeoMesa has optimized indexes for point features
and for linestring/polygon features. For linestring features,
GeoMesa supports parallel arrays of attributes for storing
information specific to each vertex of a linestring feature
(speed, heading, etc). Finally, GeoMesa natively handles
GeoJSON where you have a lot of flexibility in how you model
your data. In general, there are lots of ways to handle
rollups of sensor observations in trajectories.
With
streams of sensor or observation data, usually one either
wants to see a 'live' view (or do other online processing) or
they wish to maintain an archive of the observations for later
querying and analysis. Given a stream of AIS observations,
one may want to see where certain boats are now (perhaps to
manage a fleet). Historical analyses could answer questions
about ships pattern of life, etc.
GeoMesa
has a number of capabilities. Using the Kafka DataStore, it
is possible to show that up-to-date view of a stream. The
Kafka DataStore can be integrated with a stream processing
platform like Apache Storm or Spark Streaming to provide
online analysis.
GeoMesa
leverages big databases (Accumulo, HBase, Cassandra, etc) for
storing billions of records and providing batch analysis
options. GeoMesa has integration with Spark. For the
programmer, you can get access to RDDs of SimpleFeatures. For
an analyst, GeoMesa’s integration with Spark SQL allows for
quick data discovery and visualization.
Cheers,
Jim
On 02/15/2017 11:18 AM, Mathias Herberts wrote:
Hi,
I have seen many examples of
GeoMesa which display sequences of positions for ships coming
from AIS data.
If my understanding of how GeoMesa
indexes data is correct, each position is really a feature
with its own feature id.
This makes me wonder how a
position at a given time can be updated and how can a complete
track be deleted.
The process of storing time series
in GeoMesa seems kind of artificial and at the edge of what
the tool suite was meant for initially.
If anyone could cast some light on
those points...
_______________________________________________
geomesa-users mailing list
geomesa-users@xxxxxxxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.locationtech.org/mailman/listinfo/geomesa-users