Navigating the future of compute

Applications have become indispensable in our daily lives, shaping how we work, communicate and entertain. As they become increasingly sophisticated with the advent of new technologies such as artificial intelligence (AI) or high-performance computing, this progress has sparked many questions about the silent backbone that supports these applications: compute infrastructure. The role of compute, from commodity to enabler, and the potential of cloud computing to drive digital transformation are key considerations. However, one question stands out in its complexity: What does the future hold for compute?

Despite widespread recognition of cloud computing’s benefits over traditional on-premises solutions, full-scale migration remains a work in progress. Many businesses have capitalised on cloud-native advantages through new developments or by adapting existing applications. However, migrating legacy applications that aren’t cloud-compatible continues to be a significant hurdle. To move forward, businesses must address key considerations and tailor their strategies to align with their specific needs, ensuring harmony between applications and compute infrastructure.

Private, public or hybrid?

Although cloud-native applications are highly attractive, transitioning to the cloud often entails significant costs and risks. A complete shift to the cloud is rarely justified, but for organisations prioritising innovation and strategic advantage, it’s a necessity. Understanding the differences between on-premises infrastructure and a hybrid environment is essential for making informed decisions on cloud migration.

One major consideration is data gravity. Application performance can suffer as data volume increases and the data path grows, especially for data-intensive workloads. When these applications are hosted in the cloud but rely on on-premises data, their value can be significantly diminished due to latency in data retrieval and additional costs associated with data transfer.

Security can also present a challenge. On-premises systems traditionally rely on perimeter-based firewalls, whereas cloud environments necessitate a zero-trust model with encryption for data at rest and in motion. The shared nature of cloud services demands robust intrusion detection and multi-factor authentication to safeguard data.

Furthermore, while cloud providers now offer GPU-accelerated services to match on-premises compute power, these can be prohibitively expensive. For instance, OpenAI reported that training the GPT-3 model required US$3.2 million in computing resources alone.

Hybrid environments can also complicate inter-application communication. Migrating one application to the cloud while leaving others on-premises can increase response times, potentially leading to transaction failures or user dissatisfaction.

For businesses with extensive legacy application portfolios, migrating to the cloud involves also weighing the total cost of ownership. While Kyndryl’s 2024 Readiness Report recently found that 94% of Australian leaders say tech modernisation is a high priority, the investment required in these instances needs to be considered alongside the benefits, especially if it entails significant risks or complex re-factoring. Factors like risk, cost, innovation, security and speed of change are crucial in this decision-making process.

Assessing the enterprise cloud strategy

The above considerations are just some that may prevent large portions of an enterprise application estate being easily moved to the cloud. While some enterprises may retire or re-write this final portion eventually, the majority of organisations will likely maintain a hybrid cloud environment for the foreseeable future. Many large enterprises already have a well-defined cloud strategy; businesses must adopt a strategic approach, carefully considering factors such as application portfolio, migration strategies, and the overall alignment of their cloud initiatives with their broader business goals.

The following decision matrix outlines key attributes and remedial treatments to consider:

• Data gravity: Establish a data inventory and lifecycle management team, aim for eventual application architecture consistency, implement federated queries, or consider moving large legacy systems to cloud-adjacent data centres.
• Data sovereignty: Identify regulatory requirements based on functions, establish a security governance framework that includes on-premises and public cloud, perform a cloud risks assessment especially for business-critical applications, and understand application cloud-native compatibility.
• Talent and skills: Adopt a shift-left approach for easy adaptation to changing requirements, build automation into coding, and transform organisational culture.
• Compute intensity: Understand current on-premises usage patterns, select appropriate cloud compute types based on application functions, and consider PaaS or SaaS alternatives.
• Data consistency: Establish a data consistency strategy based on status-quo and use cases, define target cloud architecture, and document inter-application dependencies.
• Legacy hardware, programming languages and middleware: Form a dedicated team to address legacy hardware and programming languages, identify short- and long-term options, consider emulation, and evaluate refactoring or rewriting.

Organising for success

That said, cloud adoption is more than just technology. Businesses must consider factors like politics, economics, strategy, technology, environment, operations and regulations (PESTERO). It’s impossible to list all for the purposes of this article but let’s use some examples from each of the areas.

Operationally, adopting agile squads that cover the entire application stack can streamline processes. Cloud computing has shifted the responsibility for technical currency to cloud providers, enabling more precise cost allocation through usage-based billing. Product managers gain greater control over infrastructure consumption but must also focus on non-functional requirements, such as technical debt. Incorporating Site Reliability Engineers (SREs) into product teams can help address these aspects proactively. Connecting compliance efforts with security, application and infrastructure teams is vital for navigating regulatory frameworks. Meanwhile, appointing a Chief Digital Officer (CDO) can drive digital transformation and ensure alignment with business objectives. Finally, implementing sophisticated strategies for data resilience and application promotion is crucial for business continuity.

Cloud services offer numerous benefits; however, they also present challenges such as cultural shifts, operational model adjustments, security and compliance issues, economic considerations, environmental impact, and governance complexities. There is no one-size-fits-all strategy for cloud migration, but, by taking a considered and tailored approach, businesses can stay competitive and adapt to the rapidly evolving digital landscape.

Image credit: iStock.com/Just_Super