Main content

How Eutelsat OneWeb uses real-time data to ensure its satellite network is up and ready

Mark Samuels Profile picture for user Mark Samuels June 19, 2024
Summary:
The low-Earth orbit satellite company uses Snowflake technology to share data across internal teams and external partners.

globe

Eutelsat OneWeb, the low-Earth orbit satellite company, needs to monitor and manage vast amounts of data, a challenge made more complicated by the fact that the company used databases that couldn’t  scale at the speed required by the business.

Data moves around a series of connections – a user terminal, such as a smartphone, satellites and the ground network, explains Miguel Morgado, Senior Product Owner for the Performance Hub at Eutelsat Group, who adds:

That triangle creates lots of touchpoints and data. Our satellites create data at nanosecond rates. Currently, just for telemetry, we ingest and transform 77 billion rows of data daily.

The huge amount of data is just one challenge. Data analysis is complicated because the company’s satellites move at 27,000 kilometres an hour. User terminals could connect to the network from locations as disparate as an aeroplane in the sky or a boat in the Suez Canal, says Morgado:

At any point in time, we need to know which user terminal is connecting to each satellite and we have handovers between user terminals and satellites every three minutes. We need to know precisely which satellite connects to which ground terminal and the user terminal could be moving. We need to be in contact to troubleshoot and pre-empt any problems.

A further layer of complication comes from commercial agreements. The firm’s network is split into administrative regions and each distribution partner has different contracts and pricing levels. According to Morgado, this means:

All that data is computed near real-time. The data from space and our satellites must come to the hands of our engineers in a matter of seconds.

Building a data mesh

To address these needs, Eutelsat OneWeb started working with Snowflake at the end of 2020. The business knew data needed to be stored in the cloud. Following a proof-of-concept study, the firm selected Snowflake and the platform entered production in March 2021. Morgado explains:

We were impressed with their partner ecosystem experience in the telecoms area. We used the Snowflake partner DataOps.live, which ingests our data and they have extensive experience with satellites and telecoms. That cross-partner capability was important because we decided to implement a data mesh.

The data mesh went live in July 2021 and supports 37 Snowflake accounts. The tenants include data for sales, satellite telemetry, the ground network, and more. There is no data duplication across accounts and high-quality data is crucial because Eutelsat OneWeb sells data to third-party partners, such as companies analysing satellite movements to avoid collisions. Morgado says:

We test every single row of data we ingest. Data monetization was in our requirements from the beginning. The fact Snowflake has a Data Marketplace was another critical decision-making factor.

OneWeb implemented ServiceNow Connect Direct in Snowflake in 2023. All data from ServiceNow is automatically processed in Snowflake, which makes it easier to correlate incidents with outages. The company also recently created integrations with Salesforce. Morgado believes that Snowflake’s continued move towards Native Apps means it will be even easier for his company to sell data to third-party partners:

Choosing Snowflake was the right decision. At the time, we knew it was one of the best databases. As we’ve developed our capabilities, it’s been much easier to find Snowflake developers and resources than it would have been with other databases.

Monitoring performance proactively

Eutelsat OneWeb’s integrated approach to Snowflake technology has provided a range of benefits to the business and its users. OneWeb has developed a self-service approach that means anyone can get hold of the data they need, Morgado explains:

Having the information in the hands of our engineers to troubleshoot problems and predict when the issue will happen is important. Every engineer can create a dashboard. It’s simple – they connect to Snowflake and get access to the data they want. They can access this information in seconds.

Moving to a cloud-first approach and a data mesh is a significant shift for any company. Morgado’s team needed to prove the value of this move quickly:

Transforming data costs around 10 times more than ingesting it. In the first month, we could show the value of our work. We had teams that had never worked together that could suddenly exchange data and share knowledge. That success made it easier to say to management, ‘We need more investment in these areas.’

The long-term aim of the Snowflake implementation is to continue finding new ways to process data and predict potential network issues proactively. Morgado says Artificial Intelligence (AI) is likely to play an increasingly important role in this real-time monitoring activity:

Doing AI and Machine Learning is not straightforward. You need a big investment in data quality and foundations, which we have done in the last four years. That's why we feel ready to move to the next step and to do that effort securely…Shockingly, many people don't invest in data quality these days. Data monetization and sharing data will not work without data quality in place.

Loading
A grey colored placeholder image