|日本語
|
|
|
|
|
|
|
|
|
|
|
|SYSTEMS
Logs, metrics and traces: what do we see before the system breaks?
Observability is not about finding someone to blame after an error; it is about making system behaviour understandable.
記事を読むPRACTICAL / BLOG
EN — 2026考えたこと、学んだことを共有します。
カスタムソフトウェア、AI、テクニカルSEO、性能、セキュリティに関するWYNのノート。
Wynさんのメモ / 60 TEXT
技術的な問題については、人間の言葉でお話ししています。
意思決定に役立つカスタムソフトウェア、AI、サイバーセキュリティ、最新のソフトウェア発表、可視性に関するメモ。
最新の投稿を最初に表示する·4 / 6
CURRENT / AI SECURITY
Which risks does the OWASP LLM Top 10 2026 remind us of?
Protecting the model is not enough; the output, plugins, data sources and agent permissions need protection too.
記事を読むRELIABILITY
SLO, SLA and error budget: what does 99.9% really say?
Instead of presenting uptime percentage alone, tie user experience to measurable goals and technical priorities.
記事を読むPRIVACY
Privacy by design: a decision from the start, not a document afterwards
Protecting personal data is not only about writing a notice; it is about designing the product so it never needs unnecessary data.
記事を読むCURRENT / GOVERNANCE
What does NIST’s AI-assisted cyber security approach say?
Using AI in security requires increasing analysis and reporting capacity without removing human decision responsibility.
記事を読むDATA & OPERATIONS
How do you plan a zero-downtime data migration?
Moving users to a new structure without making them wait needs a controlled transition plan, not a one-off copy.
記事を読むCOOKIES
Cookie consent is not just a pop-up
Instead of bundling mandatory technical records with ad tracking in one button, what is required must be separated clearly.
記事を読むCURRENT / CLOUD
Why is following the Kubernetes CVE feed not enough?
Whether a vulnerability affects your cluster depends on configuration and components, not only on the CVE title.
記事を読むSEO & CONTENT
Content decay: when should old articles be updated?
SEO is not only producing new articles; it is regularly improving content that is no longer correct, complete, or meeting search intent.
記事を読むLOCAL ARTIFICIAL INTELLIGENCE
When is local AI the right choice?
An in-house model or local inference makes sense not for every problem, but when data control, latency and operations needs are clear.
記事を読む