Data Ingestion
Nazare can collect, store, and transform more than 10 million data points per second at the machinery. This is because Nazare is equipped with a high-speed big data processing engine independently developed based on Rust.
High-performance Big Data Processing
Through self-developed streaming technology, it shows multiple collection performance compared to big data processing technology and general databases, and collects 10 million pieces of data in real time. It is particularly optimized for collecting machinery data.
Real-time collection with just 1 device
One low cost server collects 10 million time series data per second in real time.
High performance streaming
The streaming technology developed in-house with Rust has 10 times the collection performance compared to JVM-based Spark and more than 2 times that of a regular database.
Parallel distributed processing
Performance can be improved by collecting and storing data simultaneously in parallel through Kafka and distributed pipelines.
High-efficiency Big Data Processing
It was developed in Rust, shows high memory efficiency and low CPU usage, and works without OOM using swap. Also, since it is stored with a compression efficiency of 90% or more, storage consumption is low.
High memory efficiency
Developed with Rust, it has more than 30 times better memory efficiency than JVM-based Spark during streaming processing, and does not occur in OOM using swap.
Low CPU usage
Columnar in-memory processing using Arrow shows a 3-4 times improvement in CPU usage compared to Spark.
High compression storage
Using a high-compression Parquet file format, it uses a compression efficiency of 90% or more, and uses up to 10 times less storage than a regular DB.
Various Collection Support
It supports both real-time streaming and deployment methods, and supports various data sources such as OT facilities as well as web servers and databases. It starts with one, but the capacity expands infinitely.
Live streaming and deployment
It supports real-time streaming collection such as Kafka and MQTT as well as batch collection methods using SQL.
Support for multiple data sources
It supports not only OT machineries and IoT devices, but also various data sources such as web servers and databases.
Non-stop horizontal scaling
Horizontal scaling can store tens of PB or more and does not slow down collection or query speeds over time.