A whole generation of developers has been told “cloud-first.” This column collects data, case studies, and commentary on the real economics—and traps—of public cloud rental models.

When a company that was “born on the cloud” goes “self-host first,” does that count as cloud exit? A repost of a deleted piece on the infrastructure coming-of-age for China’s internet giants.

Dec 4, 2025, Taobao, Alipay, and Xianyu all cratered. Users got charged while orders still showed “unpaid,” a carbon copy of the 2024 Double-11 fiasco.

A ClickHouse permission tweak doubled a feature file, tripped a Rust hard limit, and froze Cloudflare’s core traffic for six hours—their worst outage since 2019. Here’s the full translation plus commentary.

Founders here get asked the same question over and over: what if Alibaba builds the same thing? Alicloud RDS just launched Supabase as a managed service. Exhibit A.

AWS finally published the Oct 20 us-east-1 postmortem. I translated the key parts and added commentary on how one DNS bug toppled half the internet.

us-east-1’s DNS control plane faceplanted for 15 hours and dragged 142 AWS services—and a good chunk of the public internet—down with it. Here’s the forensic tour.

Deleting images and running away - this isn’t about commercial closed-source issues, but supply cut problems that directly destroy years of accumulated community trust.

Does bolting DuckDB onto RDS suddenly make open-source Postgres ‘trash’? Business and open source should be symbiotic. If a vendor only extracts without giving back, the community will spit it out."

A user consulted about distributed databases, but he wasn’t dealing with data bursting through server cabinet doors—rather, he’d fallen into another cloud computing pig-butchering scam.

Even trillion-dollar unicorns can be a house of cards when operating outside their core expertise.

When open source ideals meet commercial conflicts, what insights can this conflict between open source software communities and cloud vendors bring? On the importance of community boundary demarcation.

The paradigm shift brought by RDS, whether cloud databases are overpriced cafeteria meals. Quality, security, efficiency, and cost analysis, cloud exit database self-building: how to implement in practice!

Seven days after Singapore Zone C failure, availability not even reaching 8, let alone multiple 9s. But compared to data loss, availability is just a minor issue

A customer experienced an outrageous cascade of failures on cloud database last week: a high-availability PG RDS cluster went down completely - both primary and replica servers - after attempting a simple memory expansion, troubleshooting until dawn. Poor recommendations abounded during the incident, and the postmortem was equally perfunctory. I share this case study here for reference and review.

NetEase Cloud Music experienced a two-and-a-half-hour outage this afternoon. Based on circulating online clues, we can deduce that the real cause behind this incident was…

Both client and vendor failed to control blast radius, leading to this epic global security incident that will greatly benefit local-first software philosophy.

After Alibaba-Cloud’s epic global outage on Double 11, setting industry records, how should we evaluate this incident and what lessons can we learn from it?

Due to an “unprecedented configuration error,” Google Cloud mistakenly deleted trillion-RMB fund giant UniSuper’s entire cloud account, cloud environment and all off-site backups, setting a new record in cloud computing history!

The dark forest law has emerged on public cloud: Anyone who knows your S3 object storage bucket name can explode your cloud bill.

As a roundtable guest, I was invited to participate in Cloudflare’s Immerse conference in Shenzhen. During the dinner, I had in-depth discussions with Cloudflare’s APAC CMO, Greater China Technical Director, and front-line engineers about many questions of interest to netizens.

Tencent Cloud’s epic global outage after Double 11 set industry records. How should we evaluate and view this failure, and what lessons can we learn from it?

While I’ve always advocated for cloud exit, if it’s about adopting a cyber bodhisattva cloud like Cloudflare, I’m all in with both hands raised.

Luo Yonghao’s livestream first spent half an hour selling robot vacuums, then Luo himself belatedly appeared to read scripts selling “cloud computing” for forty minutes — before seamlessly transitioning to selling Colgate enzyme-free toothpaste — leaving viewers bewildered between toothpaste and cloud computing.

Alibaba-Cloud claimed major price cuts, but a detailed analysis of cloud server costs reveals that cloud computing and storage remain outrageously expensive.

Two days ago, the ninth episode of Open-Source Talks had the theme “Will DBAs Be Eliminated by Cloud?” As the host, I restrained myself from jumping into the debate throughout, so I’m writing this article to discuss this question: Will DBAs be eliminated by cloud?

Programmers are drawn to complexity like moths to flame. The more complex the system architecture diagram, the greater the intellectual masturbation high. Steadfast resistance to this behavior is a key reason for DHH’s success in cloud-free availability.

S3 is no longer “cheap” with the evolution of hardware, and other challengers such as cloudflare R2.

Alibaba-Cloud and Didi had major outages one after another. This article discusses how to move from cost-reduction jokes to real cost reduction and efficiency — what costs should we really reduce, what efficiency should we improve?

Hardware is interesting again, developments in CPUs and SSDs remain largely unnoticed by the majority of devs. A whole generation of developers is obscured by cloud hype and marketing noise.

Alibaba-Cloud’s epic global outage after Double 11 set an industry record. How should we evaluate this incident, and what lessons can we learn from it?

Alibaba-Cloud’s Double 11 offered a great deal: 2C2G3M ECS servers for ¥99/year, low price for three years. This article shows how to use this decent ECS to build your own digital homestead.

Once upon a time, “going to cloud” was almost politically correct in tech circles, but few people use hard data to analyze the trade-offs involved. I’m willing to be this skeptic: let me use hard data and personal stories to explain the traps and value of public cloud rental models.

DHH migrated their seven cloud applications from AWS to their own hardware. 2024 is the first year of full savings realization. They’re delighted to find the savings exceed initial estimates.

At the SACC 2023 FinOps session, I fiercely criticized cloud vendors. This is a transcript of my speech, introducing the ultimate FinOps concept — Cloud-Exit and its implementation path.

Public cloud margins worse than sand mining—why are pig-butchering schemes losing money? Resource-selling models heading toward price wars, open source alternatives breaking monopoly dreams! Service competitiveness gradually neutralized—where is the cloud computing industry heading? How did domestic cloud vendors make a business with 30-40% pure profit less profitable than sand mining?

SLA is a marketing tool rather than insurance. In the worst-case scenario, it’s an unavoidable loss; at best, it provides emotional comfort.

The real business model of cloud: “Cheap” EC2/S3 to attract customers, and fleece with “Expensive” EBS/RDS

I originally believed that at least in IaaS fundamentals — storage, compute, and networking — public cloud vendors could still make significant contributions. However, my personal experience with Tencent Cloud CDN shook that belief: domestic cloud vendors’ products and services are truly unbearable.

Guo Degang has a comedy routine: “Say I tell a rocket scientist, your rocket is no good, the fuel is wrong. I think it should burn wood, better yet coal, and it has to be premium coal, not washed coal. If that scientist takes me seriously, he loses.”

Cloud databases’ exorbitant markups—sometimes 10x or more—are undoubtedly a scam for users outside the applicable spectrum. But we can dig deeper: why are public clouds, especially cloud databases, like this? And based on their underlying logic, make predictions about the industry’s future.

Winter is coming, tech giants are laying off workers entering cost-reduction mode. Can cloud databases, the number one public cloud cash cow, still tell their story? The money you spend on cloud databases for one year is enough to buy several or even dozens of higher-performing servers. Are you paying an IQ tax by using cloud databases?

I recently witnessed a live cloud database drop-and-run incident. This article discusses how to handle accidentally deleted data when using PostgreSQL in production environments.

Ant Financial had a self-deprecating joke: besides regulation, only DBAs could bring down Alipay. Although DBA sounds like a profession with glorious history and dim prospects, who knows if it might become trendy again after a few terrifying major cloud database incidents?