start and end times

• G. Sfakianakis, I. Patlakas, N. Ntarmos and P. Triantafillou, "Interval indexing and querying on key-value cloud stores," Data Engineering (ICDE), 2013 IEEE 29th International Conference on, Brisbane, QLD, 2013, pp. 805-816.

- summary stats

- vacuum analyze

Move these out into separate line items as we flush them out

- Aggregate/batch simple features in iterator like we do for bin queries
- spatiotemporal compound indexes for all attribute indexes? (this would really only help with attribute equals queries)

- don't double store feature ID (once in row, once in simple feature) - would require changes to deserializationn and might complicate batching simple features

- automatically sample results for WMS queries based on expected result size

• • we have sampling already, for WMS queries sending more results than can be displayed in the map isn't necessary
  • would need to tie into query planning for expected result sizes

- use lazy deserialization in client

• • currently we deserialize the full feature in kvsToFeatures
  • instead we could use the KryoBufferSimpleFeature for lazy evaluation
  • we'd have to extend it so that it was mutable - possibly have some delegated mutable state

- currently hard-coded at four shards

- number of ranges we compute might need to be based on the shards so we don't create too many

DynamoDB

Cassandra

Jupyter/iPython/R notebook

Graphite

Metrics/graphite and predefined dashboards

- streaming metrics

You can't afford to use AWS for petabytes?

S3 Storage - store binary files, fairly static

Use case: Using weather stuff

Openstack?

What do we need additional Documentation for?

• HBase?
• Kafka?
• Streaming?
• Converters

Should we open up Confluence? What about documentation that changes or new tutorials?

Need to be able to audit all queries against a GeoServer WFS/WMS/WPS/REST/ETC

• Not just geoserver...

Streaming Dissemination

- quotas, rate limiting, geofencing, filtering

finishing it, almost done.

merge with blob store?

no schema evolution concerns

Don't transform to SFs, store native avro/json as "blobs" and index them

• Push down predicates to json/avro as jsonpath,avropath on the server side

• Support the features of vector tiles in geomesa (like skipping features)
• Decimation...server side support for, etc

• hook into writer and batch job api, stored in metadata table
• hook into query planner
• use z2,z3?
• configurable precision

• Intersection queries
• push attribute bloom filter into the iterator to do bitwise comparisons.