Perst Embedded Database Specifications

• J2SE 1.4 and higher
• J2EE 1.4 and higher
• Android (Open Handset Alliance)
• J2ME/CDC

• J2SE 1.1.*
• J2ME MIDP 2.0/CLDC 1.1

• Controlled recursive loading of objects (able to stop recursive loading for a specified class). All persistence-capable classes should implement IPersistent interface
• Persistence by reachability
• Fully transparent persistence using AOP tools such as AspectJ and JAssist
• Almost transparent persistence for .NET classes using generator of derived class overriding virtual properties of the class

• All primitive types
• Strings
• Arrays
• Enums
• .NET structs and embedded objects (objects with value semantic) in Java
• GUID, decimal and DateTime types
• BLOBs
• Raw types (objects serialized using system or custom serializers)
• Generic (parameterized) types

• User-defined class loaders
• User-defined memory allocators
• Custom serializers
• Custom full text search components (stemmer, parser, …)
• Explicit or implicit memory allocation (garbage collection)
• Abstract file interface to provide specific file implementations

• B-Tree
• T-Tree (optimized for in-memory database)
• R-Tree (spatial index)
• Patricia Trie (prefix search)
• KD-Tree (multidimensional index)
• Time series (large number of fixed size objects with timestamp)

• List
• Map
• Index (range search)
• Field index (extracts keys from object)
• Multidimensional index
• Set
• Scalable set (uses array for small set and B-Tree for large set)
• Spatial index
• Multidimensional index

• Object-oriented using various search and iteration methods of Perst collection classes
• Query-by-example (including range search) – implemented using multidimensional indexes
• JSQL – object-oriented subset of SQL
• Full text search: built-in full text search engine or integration with Lucene
• Native queries and LINQ (search predicate specified in native code)

Implementation

Shadow objects

• ACID
• No log file
• Fast recovery

• File-level locking
• Database-level locking
• Fine grain (object-level) locking

• Pessimistic (resource is locked before access)
• Optimistic (conflicts are detected at commit stage)

• Cooperative transactions
• Repeatable reads
• Serializable transactions

• Object cache (LRU, weak, strong…)
• Page pool
• In-memory database

• Leads category of Java/.NET object-oriented embedded database systems for storing and retrieving objects
• Superior performance is documented in multiple benchmarks including PolePosition (asynchronous and ACID versions) and TestIndex.

• Maximum number of objects: 2,000,000,000
• Maximum database size: 1 terabyte

• Library size from 250KB (Perst.Lite) to 500KB (Perst Java/.NET).
• Proven on mobile phones with heap size limited to less than 1MB

• High level of concurrency because of fine grain locking in pessimistic mode or use of optimistic transaction mode
• Can split data between several physical devices
• Possible to store BLOBs in separate storage locations, making object-caching more efficient

• Master-slave replication provides read-only access to primary database by multiple replicas
• Access to the same database permitted from multiple processes (JVMs)