Edge Computing and Time series database

Tech Paper

“ Discover the top DBMS suites for edge computing and take your data processing to the next level ”

Edge Computing and Time series database

Why Edge Computing?

In recent years, Edge Computing has emerged as a pivotal technological concept, transforming the way we process and manage data. This paradigm shift has gained immense significance due to the growing proliferation of Internet of Things (IoT) devices and the increasing need for real-time, localized data processing. Edge Computing refers to the practice of processing data closer to its source, right at the "edge" of the network, rather than sending all data to centralized cloud servers for analysis. This approach brings several advantages, and this is where your software, a specialized time-series database engine, plays a crucial role.

The Importance of Edge Computing:

  • Low Latency and Real-time Processing: Applications that require real-time decision-making, such as industrial automation, autonomous vehicles, and remote monitoring, cannot afford the latency involved in sending data to distant data centers. Edge Computing reduces this delay by processing data locally, enabling quick responses.
  • Bandwidth Efficiency: Transmitting large volumes of raw data to centralized cloud servers demands substantial bandwidth. Edge Computing reduces the amount of data transferred by sending only relevant, processed information, optimizing network usage and cost.
  • Privacy and Security: Edge Computing addresses concerns about data privacy and security by keeping sensitive data localized. This mitigates potential risks associated with transmitting sensitive information over public networks.
  • Offline Functionality: Edge Computing allows devices to operate autonomously even when connectivity to the cloud is lost. This is crucial for scenarios where continuous functionality is essential.
  • Scalability: With the rapid growth of IoT devices, centralized cloud processing might become a bottleneck. Edge Computing distributes the processing load, ensuring scalability without overwhelming cloud resources.
The Role of Time-Series Database Engines in Edge Computing

Time-series data, which records values over time, is prevalent in IoT applications—temperature readings, stock prices, energy consumption, etc. Time-series database engines excel at efficiently storing, retrieving, and analyzing such data.

Here's how they contribute to Edge Computing:

  • Data Storage and Retrieval: Time-series databases are optimized for high-throughput storage and retrieval of time-stamped data. This ensures quick access to historical and real-time data, crucial for decision-making at the edge.
  • Aggregation and Analysis: These databases allow you to perform complex analyses on time-series data, even in resource-constrained environments. Aggregations, downsampling, and statistical computations can provide meaningful insights.
  • Compression: With edge devices often having limited storage, time-series databases offer data compression techniques, enabling efficient storage of vast amounts of data in constrained environments.
  • Data Contextualization: Time-series databases enable contextualizing data by associating it with relevant metadata. This contextual information enhances the value of the data collected from edge devices.
  • Event Triggering: Time-series databases can support event-triggering mechanisms, enabling real-time alerts and actions based on predefined conditions, directly enhancing the responsiveness of edge applications.

The rise of Edge Computing is closely tied to the proliferation of IoT devices and the need for efficient, real-time data processing. Ability to manage time-series data efficiently aligns perfectly with the demands of Edge Computing, enabling localized processing, quick decision-making, and optimized resource usage. As industries continue to adopt and leverage Edge Computing, Time series DBMS stands at the forefront of enabling this transformative technological trend.

Edge Computing and Data Processing Requirements

How much data do you need to process at the edge, and how fast?

The clearer the answer to this question, the more likely it is that the vision of the future presented by Edge Computing will be at the customer's level.

As an extreme example, the data processing requirements for edge devices for real-time monitoring and management of NC (Numerical Control) equipment are about 1T Bytes of data every two weeks. To translate this value more precisely, it needs to digest 872,628 bytes per second.

Converting this to the average size of real-world sensor data (about 20 bytes), we calculate that it should be able to store about 400,000+ events per second.

Not only does the edge device need to ingest 400,000+ sensor data per second, but it also needs to store it in a secondary storage device for real-time analysis.

Furthermore, if the environment requires the edge device to support high availability, it will require a higher level of data processing technology.

In the end, different industries will have different types of data, different forms of data, and different amounts of data, and the requirements in various forms will become more and more stringent over time.

Machbase and Edge Computing

As a real-time time series database, Machbase already offers a standard edition for ultra-fast data processing on edge computing.

It can store 400,000 sensor data per second on a Raspberry PI 4, while utilizing CPU and disk usage very efficiently.

Even if it is an edge device, Machbase with big data technology can store data to the storage limit, and the search and analysis of big data is excellent.

Depending on your future roadmap, you can build a cluster on multiple edge devices, which means that your edge computing needs for high performance and high availability are well supported. To summarize Machbase's performance

In ingestion point of view

  • In edge Device (ARM 2 core) : up to 500,000 rec/sec
  • In single Server (INTEL 8 core) : up to 2,500,000 rec/sec
  • In Cluster (16 node) : up to 100,000,000 rec/sec

In extraction point of view

  • 0.1 seconds for extracting 10,000 random time ranges from a 400 billion data store

In compression point of view

  • 4TB store : around 1 trillion record

As you can see, Machbase already fulfills a great data processing requirement for edge computing, and if you want to evaluate the product in more detail, download and test it.

@2023 MACHBASE All rights reserved.