Valkey and the Hidden Foundations of the Digital Enterprise

Customers rarely think about the infrastructure supporting their digital experiences. A mobile subscriber expects account information to appear promptly and a retailer needs transactions to complete without disruption. Few consider the databases, caches and other software components working behind the scenes. Yet when those foundations struggle, the consequences become very visible indeed.

Much of the technology industry’s attention is directed toward AI and sophisticated digital platforms. Beneath these developments, an equally important evolution is taking place in the foundational software that makes modern applications work.

Valkey, an open-source, Redis-compatible in-memory data store, offers an interesting example. What began as a response to changes in Redis licensing has grown into a broader undertaking, with expanding technical capabilities and ambitions.

At ValkeyConf in Prague in October 2026, discussions with contributors showed progress since my first visit to KubeCon London in 2025. We explored whether abstraction distances organizations from the engineering expertise they still need. These questions matter most to communications service providers (CSPs), network infrastructure suppliers and others running complex digital services, but their implications extend across the enterprise.

Beyond the Fork: Establishing a Technology Foundation

Initially, Valkey’s significance was closely tied to licensing, compatibility and the freedom to develop an established technology without dependence on a single commercial owner. Two years later, that explanation feels incomplete.

Successive releases have improved performance, memory efficiency, observability and operational tooling, and the ecosystem has expanded to search and modular extensions. Valkey can support application caching, session storage and other tasks requiring rapid access to frequently used or short-lived data.

Its community has also broadened. The original technical steering committee comprised Amazon Web Services, Google Cloud, Huawei, Tencent, Alibaba and Ericsson, later joined by Apple and Percona. Other contributors, such as Oracle, add expertise in managed services and specialized development.

That diversity is encouraging, although resources and influence aren’t necessarily evenly distributed. The question now is whether Valkey can establish a lasting enterprise role on its own merits, rather than primarily as an alternative to Redis.

Why Foundational Software Deserves Greater Attention

Caching and in-memory data technologies rarely feature prominently in enterprise transformation discussions. Yet by holding frequently accessed information in memory, they can reduce repeated database load and improve responsiveness. Deployed well, they make the difference between a digital service that feels immediate and one that doesn’t. Their value becomes particularly apparent in environments where even modest delays can affect large numbers of transactions or interactions.

Consider a telecom operator supporting millions of subscribers. Fast access to information matters, but so does its accuracy. Cached information can become outdated, and cache failures can shift demand onto underlying systems. For charging decisions and account balances, speed can’t be separated from accuracy and resilience. Similar pressures arise in banking, retail, healthcare and public services.

When Booking.com moved its post-booking service from on-premises Redis to Amazon ElastiCache for Valkey, adding capacity to overcome initial limits still left the application slower than expected. Engineers traced delays to how it connected to the cache and handled simultaneous requests. Improving that connection reduced the time within which 99% of cache reads completed from 50 to 3 milliseconds; addressing unnecessary waits reduced the application’s slower response times by about 200 milliseconds.

The lesson was to diagnose the full application rather than prove one engine’s superiority: managed infrastructure still requires application-level engineering.

Network infrastructure suppliers share these concerns. Ericsson’s contributions to Valkey’s performance and efficiency don’t by themselves prove operational benefits in its products, but they do show foundational open-source engineering aligning with network technology requirements. Another session showed that a service can appear healthy even as backup copies fall behind. Recovery needs spare capacity as well as everyday performance.

As platforms make these capabilities easier to consume, much of the complexity shifts to technology developers, infrastructure specialists, internal platform teams and the organizations depending on their services, but it doesn’t disappear. Enterprises still need to know where responsibility sits when performance, reliability or security falls short.

Open Governance Needs a Clearer Enterprise Story

A revealing admission in Prague was that Valkey’s engineering-led community has yet to communicate a clear picture of its place in wider enterprise architecture. Redis helped establish the market for in-memory data technologies, but familiarity with that category doesn’t fully explain when, why or how organizations should adopt Valkey. Our Survey: Senior Leadership IT Investment, 2026 research found that open-source options were built into evaluations or regularly influenced purchasing decisions for 68% of respondents.

The project’s Linux Foundation governance model provides a counterweight to single-company control. There’s evidence that it works: competing contributors have collaborated on search and data-migration designs, and six-month release cycles sustain progress. Yet distributed authority doesn’t guarantee distributed maintenance effort. Growing volumes of contributions, including AI-assisted code, are already increasing the burden on reviewers.

For telecom operators and infrastructure suppliers managing software over long life cycles, source-code access remains only part of the equation. Security maintenance, predictable upgrades, technical support and accountability still need owners. A community doesn’t need to follow one supplier’s vision, but Valkey must explain its architectural fit and practical value as convincingly as its technical features.

Balancing Abstraction, Flexibility and Engineering Responsibility

Valkey modules illustrate how a common software foundation can accommodate specialized requirements without forcing developers to maintain their own version of the underlying code. An Oracle engineer working on managed caching described modules used for requirements including cloud identity integration. A shared development kit using the Rust programming language makes such extensions more accessible.

That flexibility brings obligations: modular extensions still need skilled development and testing, as a fault can disrupt the wider system. The community is also developing tools to simplify upgrades, scaling and recovery, and ways to save data with fewer interruptions or memory spikes. Both are works in progress and involve compromises, including possible delays when updating stored information.

A network infrastructure supplier might justify specialist development for particular integration needs, whereas a smaller service provider may achieve more with a supported managed service. The distinction between organizations that develop capabilities and those that primarily consume them remains important, even if the boundary is fluid. A retailer doesn’t need to become an infrastructure software supplier to compete digitally; it should understand when deeper technical involvement creates sufficient business value to warrant the responsibility.

When Development Becomes Easier, Engineering Judgement Matters More

Modules also prompted a less comfortable discussion about software engineering skills. Managed platforms and, increasingly, AI-assisted development allow engineers to achieve more without mastering every underlying component. This is progress, but how will newer developers acquire the understanding needed to diagnose failure or assess the quality of what they build?

Valkey contributors saw both sides. AI can help application developers work in unfamiliar languages, lowering barriers to contribution. Yet maintainers face more generated code requiring scrutiny, particularly when its authors cannot explain the design decisions. Project leaders described using automated reviews to manage the load but recognized that review capacity remains a constraint.

AI is also creating new uses for Valkey to cache model responses and manage an agent’s working memory. Examples at the conference highlighted the risks: reusing a superficially similar answer can produce a confident mistake, and stale or improperly scoped memory can undermine an agent’s decisions. Fast access to information is of limited value if the system can’t judge whether that information is still correct or appropriate.

AI’s productivity gains are uneven: prototyping may accelerate while verification and architectural review demand more effort. Modules can help developers deepen their understanding through practical work, but employers must invest in that expertise rather than treating automation as a substitute for engineering judgement.

Valkey’s Next Test Lies beyond the Technology

Valkey has made meaningful progress. Booking.com’s migration offers welcome production evidence, but it also shows why component-level improvements can’t stand in for measured application outcomes. The next test is whether broader deployments establish dependable performance, manageable migration, sustainable support and credible operational value across different enterprise environments.

For CSPs and other digital enterprises, the practical starting point is to test the whole service, rehearse recovery under load and establish who owns the dependencies beneath it.

Valkey’s evolution raises a wider question: which capabilities should organizations develop, which can they safely take on and what expertise must they retain? Organizations don’t need to build every layer supporting their digital services. They do, however, need to understand those foundations well enough to recognize when convenience comes at the expense of control.