1. 1. Query Language: ArangoDB uses a slightly modified version of SQL as its query language, allowing users with SQL experience to easily interact with the database. On the other hand, Dgraph uses a graph-based query language called GraphQL, which is optimized for querying graph data structures and offers more flexibility in terms of shaping the data returned.

2. 2. Data Model: ArangoDB is a multi-model database, which means it supports multiple data models including document, key-value, and graph. This allows it to handle a wide range of use cases without requiring data duplication. Dgraph, on the other hand, is a graph database and focuses solely on managing graph data structures, providing more advanced features and optimizations specific to graph data.

3. 3. Scalability and Distributed System: ArangoDB is designed to be a highly scalable distributed system, allowing horizontal scaling through sharding and replication. It provides built-in clustering and synchronization mechanisms, making it suitable for large-scale deployments. Dgraph, on the other hand, is built on a distributed architecture from the ground up, using a consensus algorithm called Raft for replicating and ensuring consistency across nodes. This makes it highly scalable and fault-tolerant, specifically tailored for distributed graph data.

4. 4. Performance Optimization: ArangoDB focuses on providing high-performance execution of complex queries by utilizing various indexing techniques, including hash indexes, persistent indexes, and geo-spatial indexes. It also offers built-in caching mechanisms to improve query response times. Dgraph, on the other hand, optimizes performance by leveraging its native graph storage and traversal mechanisms. It uses a memory-mapped file-based storage for efficient storage and retrieval of graph data. It also employs caching and prefetching techniques to minimize disk I/O and speed up query execution.

5. 5. Consistency Model: ArangoDB follows a strict ACID (Atomicity, Consistency, Isolation, Durability) consistency model, ensuring data integrity and correctness. It provides multi-document transactions and supports various isolation levels. Dgraph, on the other hand, provides a more relaxed consistency model known as eventual consistency. It favors availability and partition tolerance over strict consistency, allowing for higher scalability and fault tolerance at the expense of potential temporary inconsistencies.

6. 6. Community and Ecosystem: ArangoDB has a larger and more established community, with extensive documentation, active community forums, and a wide range of third-party integrations and tools. It also has a well-defined roadmap and regular releases. Dgraph, being a relatively newer database, has a smaller but growing community. It offers extensive documentation and actively maintains its open-source repository, but its ecosystem is still evolving.

In Summary, ArangoDB and Dgraph differ in terms of query language, data model, scalability, performance optimization, consistency model, and community/ecosystem. ArangoDB is a multi-model database with a flexible query language, while Dgraph is a specialized graph database optimized for graph data structures. ArangoDB focuses on providing high-performance execution of complex queries, while Dgraph leverages its native graph storage and traversal mechanisms. The consistency models also vary, with ArangoDB offering strict ACID consistency and Dgraph favoring eventual consistency. Finally, ArangoDB has a larger and more established community and ecosystem compared to Dgraph, although both databases have documentation and active development.