Cloud Performance Lab project for one of the world’s leading mobile operators

Together with one of the world’s leading mobile operators Aspire successfully finished the first part of the Cloud Performance Lab project within the planned timeframe.

Suboptimal Cloud performance leads to higher OpEx and CapEx, introduces network instability, and affects customer experience and/or businesses. In the era where energy is an important factor, leading Tier-1 operators are exploring ways to consume less but achieve more. Cloud operational efficiency is becoming the focus and this project was at the forefront of that for the client.

The project consisted of building and benchmarking the performance of next-gen Cloud-Native based on Kubernetes, leveraging different NIC vendors, virtualization technologies, and SmartNIC features.

Our client is planning to introduce SmartNIC features and HW offloading methods in the production network. The project identified several SmartNIC features as a potential solution, such as DDP (Dynamic Device Personalization) or ASAP (Advanced Switching and Processing) which offloads Telco workloads onto the NIC chipset. Other HW accelerators were identified, e.g. SST-BF, and RSS. The solution also provides SW GRUB & BIOS setting on servers for higher System performance and provides results on SR-IOV and DPDK efficiency.

The solutions identified throughout the project can bring various benefits for the client, such as lower CapEx/OpEx, higher network efficiency, optimal E2E vertical stack performance (Infra <> Cloud <> App), and a Cloud-native solution built in operator’s Lab (with potential reuse). Performance benefits of identified SmartNIC features are covered in next phase of the project, which is ongoing, exploring their applicability in advanced 5G low-latency and security use cases.

We congratulate the members of the project team from both sides on the amazing work, professionalism, and technical skills applied during this engagement, and wish them success in the next phase of the project!