‘Anything beyond 5ms is detrimental': How Centrica Energy migrated its most latency-sensitive applications to cloud

Interview with Senior Trading Infrastructure Change Lead at Centrica Energy

Facing growing data volumes, regulatory complexity and ultra-low latency demands, Centrica Energy migrated its most latency-sensitive applications to cloud, enabling high-performance trading, simplified operations and a foundation for global automation and resilience.

Part of the Centrica Group, Centrica Energy is a global energy trading company with operations across eight locations in Europe, Asia and North America. The organisation trades power, gas, liquefied natural gas (LNG) and renewable energy certificates across highly interconnected global energy markets.

Those markets are inherently dynamic and often volatile, placing significant demands on the technology platforms that support trading activity. Systems must combine massive capacity with extreme performance and agility. When Patrick Babic took on the role of Senior Trading Infrastructure Change Lead in 2021, Centrica Energy was struggling with the limitations of an ageing on-premises infrastructure, alongside increasingly complex regulatory and compliance requirements.

“Centrica as a group has had a cloud-first strategy in place since 2017,” Babic explains. “We were following that strategy, but our trading systems took longer to become migration ready. At that point, we still had three core trading systems running in a single on-premises data centre. Storage became a critical focus because trading applications need to be as close to storage as possible to avoid latency and performance issues.”

100 million files per hour

Centrica Energy relies heavily on its Energy Trading and Risk Management (ETRM) platform, which manages the full trade lifecycle. The system connects to live market data feeds and supports real-time risk management, compliance, reporting and analytics.

“Anything beyond five or six milliseconds of latency is detrimental to performance,” Babic says. “The environment is extremely latency sensitive. Trading systems write vast numbers of audit logs and constantly pass data back and forth. Almost every activity in a trading organisation is tracked, audited and stored almost immediately. We’re talking thousands of logs per second.”

As Centrica Energy expanded into North America, the challenge intensified. “The North American market has even more stringent requirements,” Babic adds. “There is significantly more data to store and audit.”

AWS was selected as the strategic application platform, but the choice of storage equally critical. Babic’s team adopted Amazon FSx for NetApp® ONTAP®, a fully managed, cloud-native storage service. The solution also uses NetApp® FlexCache® to position data closer to latency-sensitive applications, caching files in strategic locations to optimise performance.

The scale of the migration was substantial: 150TB of trading data and approximately 800 million files, all to be migrated within a narrow eight-hour window. It was also one of the first FSx for NetApp® ONTAP® deployments of this size in Europe.

“Migrating file-based workloads from on-premises to the cloud is challenging,” Babic acknowledges. “At peak, we were migrating close to a million files per day. Block-level migration would have been easier with replication and mirroring, but file-level migration is inherently more complex.”

The team adopted a highly structured approach. “We had to pre-stage everything and work closely with the trading teams to secure cutover windows,” Babic explains. “That meant building a custom migration solution tailored specifically to our environment.”

Most of the effort went into preparation rather than execution. “Due diligence is everything,” he says. “You need to understand your requirements in detail and set the platform up correctly from the start. I typically spend three to four months purely on preparation before touching the actual migration.”

Once configured, Babic describes the NetApp platform as straightforward to operate. “It’s essentially plug-and-play. You just need to fine-tune the configuration to meet your specific requirements. The bulk of the work is in defining the migration approach and, above all, testing. Test, test and test again to make sure everything behaves exactly as expected.”

Automating resilience

With the migration complete, the team can now focus on what Babic calls “the real work”: optimising performance and enhancing the application landscape.

“We have multiple application stacks focused on optimisation, trading and trade submission,” he says. “Around 80% of those applications now feed data directly into the FSx platform.”

The move has also removed a significant operational burden. “We still replicate to secondary cloud platforms, but we no longer have to worry about managing virtual hardware,” Babic notes. “That eliminates a huge amount of overhead. Maintenance, security, patching is all taken care of.”

Looking ahead, Babic plans to extend the FSx for NetApp ONTAP environment to additional locations, including Singapore and Denmark. His focus is now shifting towards optimisation and automation.

“We have a strong footprint across Europe,” he says. “My next priority is automating failover and disaster recovery. Managing failover at this scale is complex, so we’re using Ansible to automate regional failover between locations such as London and Frankfurt. The goal is to have a single-touch button to fail over storage and infrastructure between regions.”

For Centrica Energy, the result is a cloud-native trading platform built to handle the scale, speed and regulatory demands of modern global energy markets without the operational constraints of legacy infrastructure.