Can OSS manage comms networks, data centres and renewable energy sources?

Have you noticed that just a few people are writing about AI lately? No? None? Never seen an article about AI? Almost every article I read these days has strong undertones (or overtones) about AI. It’s almost as impossible to miss as when Sam Altman was abruptly ousted from his position by OpenAI’s board of directors in late 2023. AI was, and is, everywhere!

If everyone else is talking about AI, that makes me loathe to talk too much about it here on the PAOSS blog. There are so many other interesting OSS things to talk about too after all. Having said that, the combination of GenAI and the ease of use of ChatGPT’s interface have helped trigger one of the most powerful technology revolutions I can remember. It’s the ramifications for OSS relating to the widespread use of LLMs that we’ll talk about today.

The burgeoning demand for AI technology is causing significant challenges for companies that are (were?) striving to achieve net-zero carbon emissions. As illustrated by Google’s recent environmental report, AI’s energy consumption is pushing the boundaries of sustainability commitments. The rapid rise of LLMs was simply not foreseen when the likes of Google were setting their initial net-zero goals.

Against this backdrop of escalating energy consumption, I’ve already long been a believer that OSS can play a crucial role in the cross-over worlds of telco networks, energy (particularly renewable energy), data centre workloads, and sustainability optimisation.

Like GenAI, Digital Twins and Reality Twins have become common buzzwords too. OSS are the original Digital Twin. They’ve been a virtual representation of a system (the global telco networks) for decades. There’s no reason why our OSS can’t be the foundation of digital twins that combine networks, energy and DCs to coherently manage resource consumption / utilisation across all.

Understanding the Energy Challenge

AI, particularly generative AI, requires vast computational resources, leading to increased energy consumption. You probably already figured that part out, or knew it already. We’re already seeing Data Centres being purpose-built to satiate the demand for AI. These GPU-heavy DCs are particularly energy-intensive. As companies like Google and Microsoft expand their AI capabilities, and commit to growing their DC footprints, they also face growing power requirements that threaten their environmental goals.

This probably implies that renewable energy source projects will need to grow in unison with DC expansion, arguably making energy-DC-network investments more closely coupled than ever before.  This ever-closer coupling underscores the opportunity for innovative solutions to combine data to deliver more holistic optimisation. That is, to collect & stitch the combined data in ways that allow for optimisation of network management, data centres and renewable energy sources together.

The Role of OSS in Optimising Network Management

OSS are clearly already pivotal in managing the complex infrastructure of telecommunications networks (which includes comms rooms / DCs and power consumption BTW). The scope of OSS can therefore extend further into data centre management (DCIM) and energy optimisation. There are already signs of this happening. There are also tools with cross-over capabilities already available on the OSS market.

Our personal OSS sandpit shows just one example of how these worlds – Renewable Power, Supervisory and Comms Networks – can be combined via an OSS solution. This sandpit also includes the modelling of Data Centre and interconnecting Network infrastructure.

Here are several other ways OSS can contribute:

1. Energy-Efficient Network Design: OSS can help design networks that minimise energy consumption by optimising the layout and operation of network components. This includes selecting energy-efficient hardware and designing networks to reduce power usage during off-peak times. It may also include appropriate spacing of equipment within racks to balance out heat loads within a DC or comms room
2. Dynamic Resource Simulation and Allocation: OSS can dynamically allocate resources based on real-time demand, ensuring that servers and network components are only active when needed. This reduces unnecessary energy consumption and enhances overall efficiency. Moreover, similar solutions can also provide capacity / utilisation simulations to be more proactive with component turn-up / turn-down
3. Predictive Maintenance: By leveraging predictive analytics, OSS can foresee potential failures in network components and data centres, enabling preemptive maintenance. This reduces the associated downtime and energy waste. It may even allow for us to carry less resiliency. For example, a hot-hot / load-balanced redundancy model will generally use more energy than a cold-standby model because we don’t have duplicated infrastructure in fully operational states
4. Renewable Energy Integration: OSS can facilitate the integration of renewable energy sources into the network infrastructure. By managing the distribution and storage of renewable energy, OSS can help balance the supply and demand, reducing reliance on non-renewable energy sources. It can also potentially assist with the demand side by incentivising use during off-peak periods (or disincentivising during peaks) or fully-charged periods
5. Cooling Systems: Cooling systems are a major energy consumer in data centres. OSS can optimise by adjusting cooling based on real-time conditions, significantly reducing energy consumption
6. Real-Time Energy Monitoring: OSS can provide real-time monitoring of energy usage across networks and data centres, identifying inefficiencies and enabling prompt corrective or re-balancing actions
7. Load Balancing: By distributing workloads across multiple data centres and scheduling to align with energy supply (which tends to be less consistent night and day with renewable sources such as solar panels or wind turbines), OSS can ensure that no single facility is overburdened, reducing peak energy usage and improving overall efficiency
8. Carbon Footprint Reporting: OSS can generate detailed reports on the carbon footprint of network and data centre operations, providing insights into areas for improvement and helping companies meet regulatory and sustainability targets
9. Real Estate and Cost Modelling: Since telco and DC facilities generally reside on large land parcels, there is the potential to install renewables (eg wind and solar) locally. It can also manage energy to determine whether to use local generation or grid power supply – either from a facilities cost planning perspective (pre-installation of renewables) or a real-time utilisation perspective (post-installation of renewables)
10. Multi-utility Business Models: OSS are designed to manage networks. Many, such as the sandpit (see links above), are able to manage any type of network, whether that’s a distributed telco network, an intra-DC network, a power / water / rail / sewerage / etc network. Rather than making optimisation decisions based on the different parts of a multi-utility business, they can be consolidated into common dashboards with appropriate decision levers

Conclusion

The rapid growth of AI presents significant challenges for sustainability, but OSS offers a pathway to optimise energy management and integrate renewable energy sources effectively. By leveraging flexible technologies and innovative approaches, OSS can help tech companies meet their environmental goals while supporting the continued expansion of AI capabilities. Embracing these solutions could even been essential for balancing technological advancement with the urgency of sustainability.