DynamoDB Before and After GSI
?
1 like?850 views
Global secondary indexes (GSIs) in DynamoDB were introduced to allow querying on attributes other than the primary key. Previously, workarounds like parent-child models or separate query tables were needed. GSIs automatically manage a secondary index table that can be queried using attributes from the primary table. Writes go to both the primary table and GSIs. GSIs have their own throughput capacity and are eventually consistent. They support different projection types and allow fetching missing attributes from the primary table. The document discusses how GSIs implement a read model pattern and provide more flexible querying capabilities than previous workarounds, though they have some limitations and costs.
![[Coffee Break] CQRS
https://cqrs.?les.wordpress.com/2010/11/
cqrs_documents.pdf
? Greg Young s CQRS introduces
an architecture of separating
Read and Write models
? GSI is a small sample of
constructing read model
against a parant table](https://image.slidesharecdn.com/dynamodb-beforeandaftergsi-150319055039-conversion-gate01/85/DynamoDB-Before-and-After-GSI-9-320.jpg)
More Related Content
Similar to DynamoDB Before and After GSI (20)
More from Junya Hayashi (11)
Recently uploaded (20)
DynamoDB Before and After GSI
- 1. DynamoDB? Before and After GSI 2015-03-19? Junya Hayashi (XICA Co., Ltd.)
- 2. Before GSI
- 3. ? Hash key is ?xed on table ? Couldn t query on attributes ¡ ? So, many work arounds were invented to query on tables.
- 4. parent key model ? By using parent object id as a hash key, you can query target objects against a primary object id attributtes key tenant_id hash key user_id range key name age
- 5. query table model ? create extra table for query ? update both table on write attributtes key order_id hash key user_id created_at updated_at attributtes key user_id hash key updated_at range_key order_id created_at Order UserOrderMap
- 6. After GSI? Amazon released Global Secondary Index? on December 2013
- 7. GSI ? GSI automatically manages query table attributtes key order_id hash key user_id created_at updated_at attributtes key user_id hash key updated_at range_key order_id created_at Table GSI write RCU / WCU RCU / WCU Eventually Consistent
- 8. Index Comparison Hash? Primary Key Composite? Primary Key LSI GSI key hash hash & range hash & range? same hash key as table hash & range? any hash key unique yes yes - - write ok ok - - scan & query scan only ok ok ok attribute projection - - KEYS_ONLY, INCLUDE, ALL can fetch missing attributes from the parent table KEYS_ONLY, INCLUDE, ALL? cannot fetch missing attributes consistency strong strong strong /? eventual eventual size limit - - 10GB? per hash key - num limit - - 5 5 on the ?y - - - ok throughput - - shared with? table independent? consume WCU for sync
- 9. [Coffee Break] CQRS https://cqrs.?les.wordpress.com/2010/11/ cqrs_documents.pdf ? Greg Young s CQRS introduces an architecture of separating Read and Write models ? GSI is a small sample of constructing read model against a parant table
- 10. Conclusion ? Though there exists some limitations and extra costs, ? Secondary Indices are very useful for query model