TiDB Reviews & Product Details

What I like best about TiDB is its ability to combine the scalability of NoSQL systems with the reliability and familiarity of traditional SQL databases. It allows me to scale horizontally without changing my application logic, which is very helpful while building growing projects. The MySQL compatibility makes integration smooth, and the distributed architecture ensures high availability and strong consistency. I also appreciate how it handles large datasets efficiently while maintaining good performance. Overall, it gives me the flexibility of a modern distributed system without sacrificing the simplicity of SQL. Review collected by and hosted on G2.com.

One thing I dislike about TiDB is that the initial setup and configuration can feel complex, especially for beginners who are new to distributed databases. Understanding how different components like PD, TiKV, and TiDB servers work together takes time. In some cases, performance tuning and optimization also require deeper knowledge compared to traditional single-node databases. Additionally, documentation can feel overwhelming because there are many deployment and scaling options to choose from. Review collected by and hosted on G2.com.

I use TiDB as a database for my real-time stock analytics platform, and it resolves quite a few problems I faced with a single large instance of Postgres. TiDB performs well under heavy load while monitoring a large number of stocks. I love the MySQL compatibility, which made the migration from Postgres to TiDB easy without needing to learn a new framework. The HTAP feature eliminates the need for dedicated ETL pipelines for analytics, and the strong consistency is crucial for accurate financial transactions. I also appreciate the separation of compute versus storage, allowing me to scale them independently as needed. The initial setup for testing the migration was easy, making the transition smoother for my team. Review collected by and hosted on G2.com.

- The plan optimizer was sometimes unstable - The UI can be better - Learning curve of distributed architecture is challenging Review collected by and hosted on G2.com.

I use TiDB as the primary database for my LLM testing project, AIBenchFlow, because it stores large volumes of data efficiently. TiDB is excellent for scalability, performance, and consistency, which is crucial as my project runs thousands of tests concurrently. The MySQL compatibility was a huge plus since it made integration fast without needing to learn a new tech stack. I really like the distributed architecture, which makes scalability seamless, allowing TiDB to scale horizontally without much of my attention. The HTAP feature is fantastic because it lets me run transactions and analytics on the same system, eliminating the need for a separate pipeline just for analytics. The initial setup with TiDB Cloud was easy, and during the testing phase of the migration from MySQL, we were able to replicate the setup within a day. Review collected by and hosted on G2.com.

One area I think could be improved is the learning curve around its distributed architecture. Though the docs are solid, it can feel overwhelming at first. More beginner friendly real-world examples would help newbies like me. The most overwhelming aspect was to understand how its components - SQL layer, storage layer etc.. work together in a distributed cluster. Better diagrams and videos could be help a beginner like me. Since, my project requires real time analytics and transactions, more concrete examples on the performance tuning would be helpful. Review collected by and hosted on G2.com.

I really appreciate how flexible TiDB is. It allows our system to grow without needing big changes to the overall structure. Even as we add new features like more agents, more analytics, and more personalization, the database doesn't slow things down or become a limiting factor. Another thing that works really well is how TiDB stays out of the way. There's no need to constantly tweak settings or worry about scaling as usage changes. This kind of easy to maintain reliability is really useful when the main focus is on AI orchestration and user experience, not on managing the database itself. The initial setup was also smooth and simple to add to our current cloud system, which makes it fit well into the larger ecosystem without causing any extra work or complexity. Review collected by and hosted on G2.com.

Monitoring and performance tracking could be more user friendly. Even though the system is strong, having simpler, more focused insights would help new startups or hackathon teams learn faster. Review collected by and hosted on G2.com.

I find TiDB incredibly reliable, which is crucial for ServiceBridge when dealing with real services and financial transactions through an in-app wallet. The trustworthiness of TiDB ensures that our records and processes stay intact without any system issues. I also appreciate how seamlessly it grows without needing a complete overhaul, allowing our team to focus on enhancing the user experience rather than dealing with technical challenges. Plus, setting up the database with TiDB Cloud was simple and integrated easily with our existing tools. Review collected by and hosted on G2.com.

There are areas that could be improved. While the core system is stable, getting a good grasp of how performance behaves in a distributed SQL setup can be tricky, especially for teams used to working with traditional single node databases. Also, monitoring and performance insights could be made easier for smaller teams. Having a more intuitive way to see how queries are behaving and how the system scales would make it easier to learn and use. Review collected by and hosted on G2.com.

I like that TiDB has native support for vectors along with full compatibility with SQL. It allows us to seamlessly use semantic similarity search while managing structured candidate data and compliance processes. Its built-in vector support means we can store embeddings and conduct cosine similarity searches directly with relational data without needing to maintain separate systems. I also value TiDB's scalability and serverless approach, which helps us handle an increasing number of resume uploads and recruiter searches without infrastructure setup or maintenance. Review collected by and hosted on G2.com.

One area that could use improvement is the need for more guidance and examples that specifically focus on optimizing vector search at a large scale. Although TiDB's built-in vector search support functions well, adjusting the performance of similarity searches and choosing the right indexing methods required some trial and error during development. More hands-on documentation that's relevant to real-world applications, like resume matching or recommendation systems, would help teams learn and apply best practices more quickly. Review collected by and hosted on G2.com.

I really appreciate how TiDB handles a lot of data coming in quickly and supports the execution of analytical queries effectively. It allows us to calculate volatility metrics and trigger alerts in real-time without managing separate systems. The scalability feature is great because as more users sign up and more trading pairs are monitored, the data increases rapidly, yet TiDB's distributed structure lets the system grow smoothly without the need for manual sharding or rewriting the database structure. Also, data consistency is a big positive, ensuring that price data, calculated values, and user alert settings are all in sync and dependable. The fact that it integrates well with MySQL made the integration smooth while still providing the advantage of a scalable distributed architecture. Review collected by and hosted on G2.com.

TiDB works well for our CryptoPulse workload, but understanding how to optimize distributed queries may need more learning than with a regular single node SQL database. Some queries had to be adjusted as the amount of data grew and having better performance tips for handling time series or high frequency data would be really useful. Review collected by and hosted on G2.com.

I really like TiDB Serverless for its built-in vector search and scalable setup. The cosine similarity search allows MindVector AI to match emotion and context vectors quickly without needing an extra database, making our design simpler. The serverless model automatically handles scaling, ensuring smooth performance during real-time recommendations. Its compatibility with MySQL made connecting to our Python backend very easy. Plus, TiDB's ability to manage both structured data and vectors in one platform increases efficiency, reduces operational complexity, and helps us develop AI-powered emotional support features more reliably. Review collected by and hosted on G2.com.

One part of TiDB Serverless that could use improvement is its documentation and examples, especially those related to vector search and AI applications. As a developer working on MindVector AI, I found it difficult at first to understand how to design a good vector schema, choose the right indexing methods, and fine-tuning performance because there weren't enough simple, practical examples. More tutorials that show how to integrate with machine learning pipelines, particularly reinforcement learning workflows, would be really helpful. Also, better debugging and visualization tools for checking vector similarity queries would make it easier for developers to check and trust their results. Adding more built-in support for AI-related tasks and clearer guidelines on best practices would definitely make development faster and greatly improve the overall experience for developers. Review collected by and hosted on G2.com.

I like how TiDB lets us manage both relational data and embeddings within a single unified architecture, making development easier and reducing operational complexity. It effectively handles scalability issues, which is crucial for a telemedicine platform that deals with concurrent consultations and multilingual interactions. TiDB's distributed design ensures reliable performance as our user base grows. Its ability to merge structural healthcare data with vector-based knowledge retrieval in one scalable system is also impressive. Review collected by and hosted on G2.com.

One area that could be improved is providing clearer guidance on best practices for combining relational workloads with vector search in production settings. Although TiDB's hybrid features are strong, real world examples focused on RAG based healthcare or knowledge heavy applications would make it easier for users to get started. We had to experiment a bit to adjust how we index data and understand how the system performs when many people are using it at the same time. Having more detailed guides on optimizing performance for mixed workloads that include both SQL and vector operations would make it easier and more confident for teams to set up and use the system quickly. Review collected by and hosted on G2.com.

I really like how well TiDB works with both organized data and vector embeddings all in one distributed database. This let me handle research metadata and embedding vectors together without needing multiple systems, which made the overall architecture of InsightForge AI much simpler. I also value its strong consistency and quick query responses, which made sure the Retrieval-Augmented Generation process was reliable and accurate. Plus, TiDB's scalability helped me manage growing research data without slowing things down. Since it works well with SQL, integrating it was easy, allowing me to create a solid, efficient, and scalable backend. The SQL compatibility helped me design schemas, query research metadata, and integrate TiDB smoothly with my Node.js backend. The distributed architecture ensured high availability and consistent performance as I added more embedding data and research documents. I also used TiDB to store and retrieve vector embeddings efficiently, which allowed for accurate semantic search in my pipeline. Its ability to handle both structured queries and similarity-based searches was key for producing reliable, citation-backed research reports. Review collected by and hosted on G2.com.

One area where TiDB could improve is how easy it is to work with vector embeddings and semantic search processes. Although TiDB can store and query vector data, getting good performance for large-scale embedding searches often needs extra query tuning and changes to the database structure. Having more built-in tools for vector indexing, optimizing similarity searches, and monitoring performance would make development easier. Also, better documentation and examples for setting up Retrieval-Augmented Generational pipelines would help developers use TiDB more effectively. Improving debugging and observability when handling mixed workloads that include both SQL and vector queries would also make the overall experience better and simplify integration. Review collected by and hosted on G2.com.