joewitt99 / data-storm Goto Github PK

Schema creation and document filling responsibility should be moved to new 
class from datastorm.eventstore.otientdb.OrientEventStore class.
This will improve code readability.

Currently queries are not cached but parsed every time when we need to read 
events. Caching of prepared queries when it is applicable will improve read 
performance.

With logging entries much better to understand problems in system ))

In order to provide better performance method that indicates whether all 
instances are placed in one and only one cluster (isSingleCluster) 
should be added to the ClusterResolver class.

While using one to one mapping aggregate and cluster, we may have a performance 
bug.

Implementation of datastorm.eventstore.otientdb.OrientEventStore#readEvents 
does not sort events by sequence number.

SpringPlatformTransactionManager integration should be added.

17:18:01,233 |-WARN in ch.qos.logback.core.ConsoleAppender[CONSOLE] - This 
appender no longer admits a layout as a sub-component, set an encoder instead.
17:18:01,233 |-WARN in ch.qos.logback.core.ConsoleAppender[CONSOLE] - To ensure 
compatibility, wrapping your layout in LayoutWrappingEncoder.
17:18:01,233 |-WARN in ch.qos.logback.core.ConsoleAppender[CONSOLE] - See also 
http://logback.qos.ch/codes.html#layoutInsteadOfEncoder for details

Create weekly JMeter perfomance tests on CI - Jenkins and provide link to them.

ODatabaseDocumentProxy should be spitted by two parts:
1. Data base factory
2. Data base proxy

Unit tests for
1. ODatabaseDocumentDynamicProxyFactory
2. ODatabaseDocumentStaticProxy
should be created.

Parametrized queries will improve code readability and safety/

All code should be covered by unit tests

Currently all events for the given Aggregate are loaded without paging such 
behavior will lead to the performance problems and should be fixed.

ODatabaseDocumentDynamicProxyFactory  uses ODatabaseDocumentTx class loader 
that can (and probably will) be used as class loader of another bundle and 
classes of dependent bundle will not see proxy instance.

1. Create profile to fail build if coverage less than 95%.
2. Complete configuration to create coverage report on site phase.

SnapshotEventStore should be implemented

Currently we have orient Db cluster per aggregate instance. It will be useful 
to create pluggable Strategy that will provide cluster creation algorithm (per 
aggregate instance, per aggregate type and so on). 

Because of http://code.google.com/p/orient/issues/detail?id=263 and to decrease 
disk space consumption out of dated snapshots should be dropped.

Such movement will add implementation extensibility and testability

It seems that ODatabaseDocumentProxy#getThreadLocalDatabase method is not 
thread safe and we need to synchronize it. We need to discuss this.  

Such mode will check developer errors and will provide possible performance 
optimizations.

Usage of connection pool will improve performance for remote servers.

Axon EventStoreManagement interface should be implemented.

Create performance tests OrientDb vs. File vs. JPA vs. MongoDb .

Check for uniquenss of sequence number + aggregate type should be added for 
events with possibility to switch off this verification to improve performance.