Serverless: Promise or Panic? Uncovering the Real Benefits and Risks

2009 marked a crucial turning point for cloud computing: the concept was formally defined, widely accepted, major vendors began their strategies, Berkeley issued a seminal statement, and China released a cloud computing white paper.

2019 became another milestone when Serverless truly entered the spotlight; although the term existed earlier, 2019 is regarded as the year Serverless gained broad recognition and strategic emphasis from many providers.

From its definition through 2019 to the present, Serverless has been praised for boosting development efficiency, lowering costs, eliminating operations, and being the future of computing, yet real‑world project adoption remains limited.

The article questions what Serverless actually delivers, why the hype often exceeds reality, and what it truly means for users, developers, and the industry.

Serverless: Hope and Panic

Serverless is seen as an expectation—a hopeful goal that may become the ultimate state of cloud computing, though its final form remains uncertain.

Having worked with many vendors' Serverless products and open‑source projects, the author views Serverless as a budding paradigm whose definition is currently narrow, with an unpredictable future that we can help shape.

Transitioning from IaaS to PaaS and now to BaaS + FaaS (some even mention CaaS), Serverless offers developers new technical dividends.

While Serverless lets developers focus more on business logic and reduces operational burdens, its elasticity also brings uncertainty and confusion for many.

Indeed, the very hope it offers is accompanied by paradoxical fear.

Elastic Capability of Serverless

Different cloud providers offer varying levels of elasticity; the promised millisecond elasticity remains a vision rather than reality. Even Alibaba Cloud, with strong underlying capabilities, cannot achieve true millisecond elasticity and resorts to instance reservation to balance load.

Second‑tier providers lag behind in capability, often relying on high costs for optimization, which is unsustainable. The author’s previous testing of cold‑start performance showed mixed results, though Alibaba and Huawei have made notable efforts in this area.

Two Common Myths About Serverless

Vendors aggressively promote Serverless, and the author aims to present an unbiased view.

Myth 1: Extreme Elasticity, Millisecond Startup, Eliminating Cold Starts

Alibaba Cloud markets its Serverless product as “extreme elasticity,” but the definition of “extreme” should be user‑driven, not vendor‑driven, making the claim confusing.

While some vendors tout sub‑second elasticity, real cold‑start times often span several seconds, revealing a gap between marketing and reality.

Describing Serverless with millisecond metrics is misleading because cold‑start latency depends on code package size and configuration; transparent data is needed instead of vague promises.

Myth 2: Ultra‑Fast Deployment

Some providers hype “instant deployment,” yet the actual speed varies with code size, network conditions, and hardware; a 1‑second deployment may be impressive, but a 300‑millisecond instance startup can be far more impactful.

Fundamentally, cloud vendors should prioritize security, stability, and performance over superficial speed claims; prolonged cold starts can render rapid deployment benefits moot.

In practice, Alibaba’s tooling is lacking, Baidu offers limited community support, while Tencent provides a decent experience with Serverless Framework integration. Nonetheless, vendors should focus more on security, performance, and stability—areas where Alibaba and Huawei already excel.

What Does Serverless Actually Bring?

From a developer’s perspective, Serverless offers hope, a liberation of thinking, and a novel problem‑solving approach.

Adopting Serverless for new services can improve engineering efficiency and operational simplicity, but migrating existing workloads is risky and often painful; support for container images would be a valuable advancement.

Serverless reshapes our understanding of cloud computing, with definitions ranging from pure FaaS to BaaS + FaaS, and even proposals to include CaaS. Regardless of terminology, it represents a new paradigm poised to become the default computing model.

Domestic Serverless development has been slow; the author hopes for a provider that combines Alibaba’s technical depth with Tencent’s user experience, ushering in an “All Serverless” future.