From raw data dump to actionable insights
Our goal at Marlinks is to turn data into actionable insights. In a nutshell, this means as a starting point we verify and enhance the quality of the raw data and afterwards we process the data to provide actionable insights to the client. Today we talk with our IT engineer Sander Van De Moortel, who’s working daily on developing and improving our software.
Hi Sander, can you tell us a bit more about who you are and what you do?
In the past, I was a linguist with a passion for programming. Now I am a programmer, who’s still interested in languages. I was educated as a linguist and after my studies I worked a few years in the world of communication and marketing. However I decided to re-educate myself in my free time to IT, and that’s how I quickly found my way into the world of programming, which I enjoy immensely.
After having spent a few years working in Indonesia, China, Taiwan and Germany, I resolutely decided to fully dedicate myself to a career in IT. On my return to Belgium, I started working in a data processing company where I could further develop my expertise in IT. After a pleasant period I was in need of a new challenge, which I have found with Marlinks. My responsibilities at Marlinks are very broad: from internal to external IT infrastructure and the integration of Marlinks services into the existing software.
How do we collect data from clients?
Together with our clients we define as a first step how to retrieve the necessary data from their premises. This could be simply done by sharing data through files or databases or it can be collected through for example a web interface.
In the second step we check if there is any historical data available. If we have historical data, our model can be operational within a maximum of three months. Without historical data, we will start collecting data with Marlinks. After six months of data collection, the model is operational.
Finally, we check whether the available data is reliable. Often we need to recalibrate the hardware as to provide accurate data.
How much data are we talking about? Is this data stored by the customer or at Marlinks?
It depends on what device we are talking about. Are we speaking about a DAS or DTS device? The amount of data that needs to be stored each day with a DAS device depends on the length of the cable, the frequency or the signal and the spatial resolution that is set. For a 6km cable and a 2000Hz signal, the amount of data to be stored each day will be around 190GB. The storage of data is huge with a DAS, therefore it is important to delete data in time or even better to only store data that meets a certain pattern, for example a scraping pattern of a CPS. For a DAS device it is also important to process the data in real-time.
With a DTS device, we do not need to know in real-time if – for example – the cable is exposed. The data is therefore less huge than with a DAS measurement. Currently, we import data every five minutes.
We can monitor the data in three different ways. This depends very much on what the customer wants.
- Marlinks manages the installation and data in the cloud via virtual servers.
- The customer provides a server and manages the data himself.
- Marlinks provides a bare metal server and places hardware on site.
Our preference goes to having a virtual server installed at the clients’ premises, since it will exclude the need for hardware maintenance or manual back-ups.
How do we ensure data security and how long do we keep the data?
We usually keep the data for five years, because the data is less relevant after more than five years. For the data of the temperature in the cable, it is useful to keep this for one or two years in order to be able to trace back the influences of the seasons. The period can of course always be discussed with the customer.
As for data security and back ups of the relevant data, ownership lies at the administrator of the server. In case Marlinks manages the server, we take care of data security and back ups.
How do we go from raw data to actionable insights?
All data that we obtain from the customer and from the DTS or DAS device is called raw data. This is simply a dump of unstructured, raw data in which there is no structure. The first thing we do is extract the raw dump of data and identify the important information that we need to develop our software. This is the information gathering phase. In the next step, the insights phase, we look for links or a structure in this data that, in the case of DAS, may point to an event such as CPS abrasion or an impact event that we already have in our library. When we have found certain events in our processed data, we can draw further conclusions about them. For example that this is a risky event that needs further investigation by the operator.
Example of data collection
We use mainly electrical and temperature data for our DoB model. For example voltage, reactive power and active power. With this data we can – for example – determine how hot it is inside a cable or how deep it is buried under the seabed. It is also important that we know what materials the cable is made of and what the effective temperature of the fiber is. All the data together has a direct influence on the temperature.
How does the Marlinks dashboard work?
The dashboard allows the customer to look at all insights individually and is very user-friendly. The dashboard summarises everything into simple graphs that are easy to interpret, for example, you can see where and when something is happening along your subsea cable. We have included some additional add ons, an example is that you can set certain parameters to warn you if your cable is not buried deep enough. Marlinks also provides a periodic report with actionable insights.
Thanks for the insights, Sander! Would you like to learn more about how we collect, manage and turn data into actionable insights with Marlinks? Get in touch with firstname.lastname@example.org.