1. Data Model: Scylla is a NoSQL database that utilizes a column-oriented data model, similar to Apache Cassandra. It allows for flexible schema design and is optimized for write-heavy workloads. On the other hand, TiDB is a NewSQL database that supports both SQL and NoSQL interfaces. It utilizes a row-oriented data model and provides strong consistency.

2. Consistency Model: Scylla follows an eventual consistency model, which means that updates to the database may take some time to propagate across all nodes. It prioritizes availability and partition tolerance over immediate consistency. In contrast, TiDB offers strong consistency, ensuring that all reads and writes are immediately visible to all clients. It follows the strict linearizability model for global consistency.

3. Replication: Scylla uses a masterless replication model called "peer-to-peer" replication, where each node acts as both a replica and a coordinator. This approach allows for high availability as all replicas can serve read and write requests. TiDB, on the other hand, adopts a master-slave replication model, where a single master node coordinates write operations and multiple slave nodes replicate the data. This ensures consistency but may introduce a potential bottleneck at the master node.

4. Scaling: Scylla employs automatic sharding to evenly distribute data across multiple nodes, enabling it to horizontally scale read and write operations. It provides automatic data partitioning based on the primary key. In contrast, TiDB uses a technique called "horizontal partitioning" to distribute data across multiple nodes, allowing it to scale horizontally. It divides tables into multiple partitions based on predefined rules.

5. Data Placement: Scylla utilizes a distributed hash table (DHT) to determine the placement of data across nodes. It uses a consistent hashing algorithm to evenly distribute data, ensuring efficient load balancing. TiDB, on the other hand, uses a range-based partitioning scheme to determine data placement. It divides data into ranges based on the primary key, allowing better locality and improved performance for certain workloads.

6. SQL Support: While both Scylla and TiDB support SQL queries, TiDB provides a more extensive SQL feature set and better compatibility with existing SQL ecosystems. It fully supports ANSI SQL and provides online schema changes, which allows altering schema and adjusting table structures without interrupting ongoing queries.

In Summary, Scylla and TiDB differ in their data models, consistency models, replication approaches, scaling methods, data placement strategies, and SQL support, making them suitable for different use cases and workloads.