Week1 - Week2 Plan¶

Prerequisite Knowledge¶

• RocksDB Column Family: Purpose and Usage
• Understanding the role of Column Families in logically and physically partitioning data within a single RocksDB instance.
• NebulaGraph's Raft-WAL Workflow
• How key-value pairs representing graph data are written to RocksDB.
• How the Write-Ahead Log (WAL) is integrated with the Raft consensus algorithm to create a replicated state machine.
• The process for WAL file deletion: when and how old log files are truncated.
• Cooperation Between Graphd and Storaged Executors
• How the query plan executors in the graphd (query) layer interact with the local executors in the storaged (storage) layer to fulfill a request.
• Folly's Future/Promise Asynchronous Programming Model
• Understanding the asynchronous programming paradigm using folly::Future and folly::Promise.
• NebulaGraph's Memory Management Approach
• How NebulaGraph manages memory allocation and deallocation efficiently, potentially using techniques like arena allocators and memory tracking.
• Thrift: Serialization, Code Generation, and Usage
• defining data structures and services in the IDL
• generating language-specific code.

Support vector in type system¶

• Implement vector value type and property type
• Implement vector value type in type system.
• Implement vector property type:
  • data encoding with column family
    
  • RocksEngine for Multi Column Family
• Implement serialization and deserialization for vector type
• Graphd to Storaged by thrift
• Graphd to Client by nebula-go

Support distance functions for vector type¶

• Support euclidean/cosine/inner_product functions in function manager
• Support k-NN search
• Test by original nGQL sentence

Support partial DDL for vector type¶

• Create tag with vector type properties

Support DML for vector type¶

• Insert vertex or edge with vector property type

Support original search for vector type¶

• Support MATCH, GO, FETCH statement for vector type