- August 12, 2013
By Bill Lydon, Editor
To meet their customers' requirements, Sabisu has developed Shika, a lightweight historian, to process, manipulate and store data at higher speeds than can be achieved with SQL databases. The customers’ challenge is to capture the data, read/write data and perform high speed calculations. These tasks require fast database interaction. I recently discussed Shika, currently in beta testing stage, with Sabisu’s Tim Sharpe. Sabisu is a software company based in the United Kingdom that has developed innovative software to empower users to build information dashboards. (See Empowering Users to Create Information Dashboards.)
Based on our discussion, Shika is a high speed non-rational database historian that uses simple addressing schema for performance. Shika has a lightweight footprint of 500 MB (4GB recommended). Shika consists of the Shika Core database and Shika Engine.
Acquiring data from other systems and devices is accomplished using OPC UA and other interfaces are planned.
Shika is designed for rapid storage and retrieval of time series numeric data and for storing large amounts of high frequency data from a wide range of sources. The goal is continuous storage of chronological manufacturing process data. Shika employs a tag-based structure familiar to automation and manufacturing IT teams.
Performance
In real-world tests, Shika has recorded read/write speeds approximately 1,000 times faster than SQL. To measure execution time, the test illustrated a representative test of a demanding real-world implementation, writing multiple values non-sequentially of 40,000 tags.
This test shows the overhead involved with Shika calling over a network using Windows Communication Framework (WCF). Even with this overhead, Shika is still much faster than a local SQL implementation. Having Shika local in a computer offers the best performance. As the table shows, the first pass to Shika takes more time, but subsequent passes are faster.
Another benchmark was done on an Intel Core2 Duo E7500 2.93GHz with 4GB RAM and illustrated higher raw data handling capacity than SQL, measuring writes per second.
Self-contained Small Footprint
Shika is self-contained making it easy to install. It can be run remotely as a server or embedded into an application. Shika has no dependencies on 3rd party database systems, special hardware nor does it require specialist skills. To illustrate the small footprint and portability, the company implemented it on a Raspberry Pi. (http://www.raspberrypi.org)
Shika running on Raspberry Pi Hardware
The Raspberry Pi is a credit-card sized ARM 11 computer that measures 85.60mm x 56mm x 21mm (with a little overlap for the SD card and connectors which project over the edges). It weighs 45g. In this application Raspberry Pi and Shika were running on Linux. I asked if a Linux version is a planned product and it is not at this time.
Getting to Product
The company currently has three beta partners implementing Shika. Sharpe discussed their interest in learning how it compares to other established process historians. Sabisu believes Shika starts up in a fraction of the time as compared to other process historians and it can beat others on raw data acquisition and access times.
Thoughts & Observations
Shika is a very interesting product that fits in the new automation model of leveraging data and analytics to improve production efficiency. Performance is important when dealing with large volumes of data where calculations are done and new values are written back into the database. The fact that Shika is so fast means users will likely find more applications. Shika potentially has many more applications that customers may have never thought about in the past.