Database Software Resources

I'm looking into which time series databases providers practitioners in similar, data-intensive industries actually trust once a system is running in production and can't afford to fail quietly, rather than which ones just score well in a general feature comparison.

• TDengine (4.7 stars, 15 reviews) earns trust specifically in industrial IoT contexts, with reviewers in manufacturing and hardware pointing to its purpose-built design for handling billions of sensor readings per day as the reason they moved away from more general-purpose databases for this workload.
• Aerospike (4.4 stars, 83 reviews) has earned trust at genuinely enormous scale, with one team in the packaging industry using it to track nearly trillions of records in real time across a global infrastructure, crediting its low administrative overhead for large clusters as what made that scale manageable without a proportionally larger operations team.
• InfluxDB (4.5 stars, 102 reviews) shows up across a wide range of industries specifically for IoT and application monitoring use cases, with its 300-plus Telegraf plugins cited as reducing the integration work needed to start collecting data from an existing tech stack.

Trust in this category seems to track fairly closely with whichever industry's specific scale and reliability demands a database was originally built to survive.

Would it be fair to say the "best" choice here depends more on matching a database's origin use case to your own industry than on overall review scores? And for anyone in manufacturing or industrial settings specifically, has TDengine's smaller community been a real obstacle when troubleshooting something unusual?

Hi G2 users, a few names keep coming up specifically for time series databases that hold up once a workload gets genuinely large and complicated, rather than the toy-scale demo every vendor shows off. I've been comparing notes across a handful of platforms with that specific angle in mind: not just storing timestamped data, but handling millions of data points a second without falling over.

1. InfluxDB (4.5 stars, 102 reviews) handles high volumes of telemetry and sensor data reliably enough that one team used it specifically to keep monitoring and metrics off their main database entirely, crediting it with making historical trend data accessible in a single click rather than requiring a separate query process.
2. QuestDB (4.8 stars, 35 reviews) was described by one team that migrated off both Elasticsearch and TimescaleDB as a genuine turning point once their ingestion needs hit roughly 3 million rows per second, with query times dropping from as long as 30 seconds down to under 500 milliseconds for the same workload.
3. Prometheus (4.5 stars, 61 reviews) is frequently paired with Grafana for visualizing metrics, and one engineer specifically highlighted being able to calculate error rates and saturation levels across thousands of containers in real time as the kind of complex, dynamic-environment workflow it was built for.

Between these three, the scale numbers people report vary wildly depending on the use case, which makes me wonder how much of "handles complex workflows at scale" really comes down to matching the database's ingestion architecture to your specific data shape rather than raw performance alone.

For anyone who has pushed one of these to genuinely high ingestion rates, at what point did you start hitting friction, and was it the database itself or more the surrounding infrastructure that needed rethinking?

Looking for input from G2 reviewers who are software engineers, backend developers, and full-stack developers in the Relational Databases category, specifically from those who have implemented a relational database as the primary data store for a production application.

The platforms with the strongest software engineer production reliability evidence:

• PostgreSQL: The WAL-based crash recovery model is specifically credited for producing databases that recover to a consistent state after unexpected shutdowns rather than requiring manual repair or data loss assessment.
• MySQL: The depth of operational knowledge available from the community, the hosting ecosystem, and the breadth of managed services built around MySQL means that production issues are reliably diagnosable and recoverable without specialised database expertise.
• Microsoft SQL Server: The comprehensive SQL Server Agent for scheduled job management and the built-in monitoring through SQL Server Management Studio are credited for giving software engineers the production observability they need to identify issues before they become failures.
• Amazon RDS: The automatic minor version patching and security updates remove the maintenance burden that software engineers running self-managed databases must schedule around active application traffic.
• SQLite: Software engineers credit the simplicity of the reliability model for embedded and edge applications where the absence of a separate database process to monitor and restart is the reliability advantage over server-based alternatives.

For software engineers who have operated a relational database in production for more than one year: what was the first production incident that was caused by database behaviour rather than application code, and was the root cause a schema design decision, a missing index, a configuration default, or a platform-specific behaviour you had not encountered in development?