Crece con confianza. Desde la implementación de un pequeño nodo de desarrollo para la búsqueda de aplicaciones hasta la gestión de una gran implementación de cientos de nodos, nuestros expertos en Elastic te contarán todo lo que necesites saber.
7. 7
Prevent Mapping Explosions
Elasticsearch Mapping
• "dynamic": true
‒ Newly detected fields are added to the mapping. (default)
• "dynamic": false
‒ Newly detected fields are silently ignored.
• "dynamic": strict
‒ If new fields are detected, an exception is thrown and the document is rejected. New fields
must be explicitly added to the mapping.
https://www.elastic.co/guide/en/elasticsearch/reference/master/dynamic.html
10. 10
Sharding Tips
• How many shards should I have?
‒ Well, it depends.
‒ But no more than 25 per GB of Java heap space on the data node.
• How big should a shard be?
‒ Well, it depends.
‒ But between 1-10 GB for a search-heavy use case, and 10-50 GB for a time series use
case.
‒ We are changing the default # of shards from 5 to 1 in Elasticsearch 7.0
For clusters of any size
• Shards take up CPU, RAM
• Segments take up CPU, RAM, File Handles
13. Rollups
Automatically roll up data into coarser
time buckets as it ages
- Save on storage space & costs
- Smaller indices = faster analytics
Manage rollup jobs in UI
Visualize rolled up indices
Aggregation functions:
- Avg, min, max, sum, count
23. Frozen Indices
Enable higher storage:memory ratio
Trades off search speeds for lower
memory footprint (i.e. lower costs)
Keep data searchable (online) in an cost-
efficient way
Operationally much simpler than
alternatives like snapshots or archival.
Data
Rollups
(6.3)
Hot / Warm
Snapshots
Minimal
Snapshot
(6.4)
Index
Lifecycle
Mgmt
(6.6)
Frozen
Indices
(6.6)
27. 27
"Here, Hold My Beer."
Customer Stories to
Learn from As Told by
Support
https://www.elastic.co/elasticon/conf/2018/sf/here-hold-my-
beer-customer-stories-to-learn-from-as-told-by-support
28. 28
Additional Resources
• Scaling Uber’s Elasticsearch cluster
‒ Danny Yuan about how Uber achieve:
‒ 850k to 1.3M messages/second
‒ Ingest 12TB / day
‒ Data Size: 1PB
‒ Link: https://www.infoq.com/presentations/uber-elasticsearch-clusters
• Elastic Discuss forum:
‒ Link: https://discuss.elastic.co/