DolphinDB delivers high-throughput ingestion, millisecond-level real-time analytics, and distributed time-series analysis for Industrial IoT. It addresses the common limitation of traditional time-series databases that prioritize storage over computation. It is well suited for high-frequency data collection, real-time alerting, historical analysis, and predictive maintenance. Keywords: DolphinDB, Industrial IoT, time-series database.
Technical specifications provide a quick snapshot
| Parameter | Details |
|---|---|
| Core product | DolphinDB |
| Primary positioning | Distributed time-series database and real-time analytics platform |
| Typical protocol scenarios | MQTT, OPC-UA, SQL/API access |
| Language interfaces | Python, Java, JavaScript |
| Architectural characteristics | Columnar storage, unified streaming and batch processing, distributed architecture |
| Core dependencies | Built-in stream computing engine, analytics function library, partitioning mechanisms |
| GitHub stars | Not provided in the source material |
| Applicable scale | TB- to PB-scale industrial time-series data |
Industrial IoT data platforms must solve both storage and computation
Industrial field data is not just “large in volume.” It is high-frequency, continuous, concurrent across many devices, and highly latency-sensitive. A database that only persists data to disk cannot effectively support real-time monitoring, anomaly detection, or cross-device correlation analysis.
DolphinDB stands out because it brings time-series storage, stream processing, batch analytics, and SQL queries together on one platform. For Industrial IoT, that means lower data movement costs and a shorter path from ingestion to decision-making.
Mainstream time-series databases have clear capability boundaries
| Type | Representative products | Characteristics |
|---|---|---|
| Specialized time-series databases | InfluxDB, KairosDB, TDengine | Emphasize write performance and compression |
| Relational extensions | TimescaleDB, ClickHouse | Balance SQL support and analytics |
| Distributed time-series platforms | DolphinDB, Druid | Emphasize scalability and computational power |
This classification shows that DolphinDB is closer to a platform-oriented time-series database than to a simple ingestion-and-storage tool.
DolphinDB’s core strengths appear in ingestion, real-time performance, and analytical depth
DolphinDB fits IIoT workloads because it does not limit database capability to the storage layer. Instead, it treats real-time computation as a first-class design goal.
Write performance determines the upper limit of device connectivity
The source material shows that DolphinDB can achieve over 1 million TPS on a single node and over 10 million TPS in distributed writes. This level of performance fits high-frequency sensors, PLC measurement points, and production-line telemetry scenarios.
| Database | Single-node write TPS | Distributed write TPS | Key reason |
|---|---|---|---|
| DolphinDB | 1,000,000+ | 10,000,000+ | Columnar storage + batch writes |
| InfluxDB | 500,000 | Not supported | TSM engine |
| TimescaleDB | 200,000 | 1,000,000 | Built on PostgreSQL |
| TDengine | 1,000,000 | 5,000,000 | Super table design |
# Pseudocode: batch write sensor data to DolphinDB
records = collect_sensor_batch() # Collect device data in batches
if len(records) > 0:
conn.run("tableInsert{loadTable('dfs://iot','sensor_data')}", records) # Submit in batches to reduce I/O operationsThis example shows the basic pattern for improving throughput in high-frequency ingestion scenarios through batch writes.
Real-time computation determines whether alerts are fast enough
In industrial scenarios, the key challenge is not only to “write the data,” but also to “compute on it immediately after ingestion.” DolphinDB includes native stream tables, a time-series engine, a cross-sectional engine, and an anomaly detection engine, which reduces reliance on external stream processing components.
| Capability | DolphinDB | InfluxDB | TimescaleDB |
|---|---|---|---|
| Stream tables | ✅ Native support | ❌ | ❌ |
| Real-time aggregation | ✅ Millisecond-level | ⚠️ Second-level | ⚠️ Second-level |
| Anomaly detection | ✅ Built in | ⚠️ Requires Kapacitor | ❌ |
| Response latency | <10ms | 100-1000ms | >1s |
A rich function library reduces the cost of industrial analytics
The source material states that DolphinDB includes more than 1,000 built-in analytical functions covering statistics, time-series analysis, aggregation, machine learning, and signal processing. For vibration analysis, condition monitoring, and quality fluctuation tracking, these built-in capabilities can significantly reduce external ETL work and custom algorithm assembly.
# Pseudocode: calculate a rolling mean and detect anomalous fluctuations
data = conn.run("select ts, value from sensor_data where device_id='d1'")
data["mavg_10"] = moving_avg(data["value"], 10) # Calculate the 10-point rolling mean
alerts = data[abs(data["value"] - data["mavg_10"]) > 3] # Mark deviations beyond the threshold as anomaliesThis example highlights that in DolphinDB-style scenarios, the priority is not just storing data, but also completing feature extraction and anomaly detection quickly.
Partitioning and distributed architecture directly affect long-term scalability
The hardest problems in industrial data platforms often appear only after several months: more devices come online, retention periods become longer, and query dimensions grow more complex. At that point, the partitioning strategy determines whether the system can still scale smoothly.
DolphinDB supports VALUE, RANGE, HASH, and COMPO partitioning. It is especially well suited for combined partitioning by time and device. This is more effective for industrial multi-device queries than pure time-based partitioning.
| Database | Partition types | Composite partitioning | Dynamic partitioning |
|---|---|---|---|
| DolphinDB | VALUE/RANGE/HASH/COMPO | ✅ | ✅ |
| InfluxDB | Time-based sharding | ❌ | ✅ |
| TimescaleDB | Time + space | ✅ | ✅ |
| TDengine | Time + tags | ✅ | ✅ |
Distributed deployment makes historical analytics and real-time workloads coexist
The architecture described in the source material includes a Controller, multiple Agents, multiple DataNodes, and distributed tables. That means the control plane, scheduling plane, and data plane can scale together, making the platform suitable for gradual evolution from a single plant deployment to an enterprise-wide rollout.
Performance and cost data suggest DolphinDB is a stronger fit for high-frequency IIoT workloads
In the referenced test environment, DolphinDB reached 1,200,000 write TPS with 5 ms write latency. It also delivered 1 ms point queries, 10 ms one-hour range queries, and 20 ms aggregate queries. These metrics are especially important for monitoring dashboards, device traceability, and process analytics.
| Metric | DolphinDB | InfluxDB | TimescaleDB | TDengine |
|---|---|---|---|---|
| Write TPS | 1,200,000 | 500,000 | 200,000 | 1,000,000 |
| Write latency | 5ms | 10ms | 20ms | 6ms |
| Point query | 1ms | 5ms | 10ms | 2ms |
| Aggregate query | 20ms | 100ms | 200ms | 30ms |
In terms of storage efficiency, DolphinDB compresses 100 GB of raw data to about 10 GB with a 10:1 compression ratio. Although TDengine performs better in this sample set, DolphinDB offers a more balanced tradeoff across query capability and computational depth.
Cost includes more than software licensing
Total cost includes not only licensing, but also server count and operational complexity. For a deployment with 1,000 devices, 10 Hz sampling, and one year of retention, the source material estimates that DolphinDB requires about 3 TB of compressed storage, 3 servers, and roughly 150,000 RMB in hardware cost. By comparison, InfluxDB requires 6 servers and TimescaleDB requires 10 servers, making DolphinDB more resource-efficient in this scenario.
DolphinDB is a better fit for industrial workloads that require real-time decisions
If your workload involves high-frequency ingestion, millisecond-level alerts, complex historical traceability, or predictive maintenance, DolphinDB’s integrated architecture offers stronger advantages than a storage-only time-series database. Its platform value becomes even clearer when you need JOINs, multidimensional aggregation, and online feature engineering.
However, if you only manage a small number of devices, collect data at low frequency, and need a basic monitoring dashboard, a lighter solution such as InfluxDB may be a better fit. If transactional consistency is the main requirement, a relational database such as PostgreSQL should take priority.
You can evaluate the choice quickly with four questions
- Does device ingestion frequency reach per-second or even millisecond levels?
- Do you require built-in stream processing and real-time alerting?
- Do you need complex SQL, JOINs, and historical analytics on the same platform?
- Do you want to reduce external components and operational coupling?
If the answer is “yes” to any two of the first three questions, DolphinDB usually deserves a top position on your shortlist.
FAQ provides structured answers
Q: Why does DolphinDB have stronger advantages in Industrial IoT than traditional time-series databases?
A: Because it offers not only high-throughput time-series ingestion, but also built-in stream processing, analytical functions, and distributed architecture. It can cover ingestion, computation, querying, and alerting within one system, which reduces integration cost and system complexity.
Q: Which industrial scenarios are the best fit for DolphinDB?
A: It is best suited for high-frequency device data collection, real-time monitoring and alerting, massive historical analytics, predictive maintenance, and multi-source data fusion, especially in systems that require both millisecond-level response and complex computation.
Q: When should you not prioritize DolphinDB?
A: If the workload is small, ingestion frequency is low, only basic monitoring is needed, or strong transaction processing matters more than time-series analytics, a lightweight time-series database or a traditional relational database is usually more cost-effective.
Core summary captures the decision-making value
This article examined DolphinDB’s core advantages in Industrial IoT across five dimensions: write performance, real-time computation, analytical functions, partitioning architecture, and cost. It also compared DolphinDB with InfluxDB, TimescaleDB, and TDengine to help developers make a better-informed time-series database selection decision.