Notes from the racks
Cloud Insights for teams running production workloads
Cloud migration playbooks, managed Kubernetes practice, infrastructure cost control, backup and disaster recovery, and EU data residency — written by the engineers who operate the Antyxsoft platform, not by a content agency.
Written by platform engineers
Field-tested, not theoretical
New analysis every month
Latest articles
34 articles
Networking/
5 min read
Connecting an office to your VPC over site-to-site VPN
Tunnels, routes and the two settings that cause most failed hybrid setups — a practical order of work for joining an office network to a cloud VPC.
Sep 3, 2026
Networking/
5 min read
Segmenting a multi-tier application the auditors will accept
Which tiers to separate, what to log, and how to document network boundaries so segmentation stands up to an audit rather than just looking tidy.
Sep 3, 2026
Storage/
5 min read
Moving 40 TB into object storage without a maintenance window
Two sync passes, a checksum verification and a DNS flip — the migration order that moves a large bucket with no downtime and no lost writes.
Sep 3, 2026
Storage/
5 min read
Serving a video library from object storage and a CDN
Where to cache, what to sign and how to keep origin traffic predictable when a growing video catalogue is served from object storage.
Sep 3, 2026
Storage/
5 min read
Designing a retention policy that survives an audit
How long to keep backups and logs, what to place beyond reach of an administrator, and what longer retention actually costs.
Sep 3, 2026
Storage/
5 min read
Growing a database volume without downtime
Resize the volume, extend the partition, grow the filesystem, verify — the order that avoids an outage, and the two steps people skip.
Sep 2, 2026
Storage/
5 min read
A snapshot strategy that survives a real incident
How often to snapshot, how long to keep them, what snapshots do not protect you from, and when to reach for Veeam instead.
Sep 2, 2026
Storage/
5 min read
Block storage or object storage: which one for this data?
A decision guide based on access pattern, latency tolerance and how the data grows — with the three cases where the answer is both.
Sep 2, 2026
How-to/
5 min read
Inference is not training: sizing for serving, not for the run
Training is throughput-bound and batch-friendly. Serving is latency-bound and bursty. Sizing one like the other is how GPU bills get strange.
Sep 2, 2026
compliance/
5 min read
Fine-tuning an open-weight model in one EU region
Keeping a fine-tune in one jurisdiction is an operations problem, not a legal one. The seven places training data actually leaves — and how to close them.
Sep 2, 2026
Infrastructure notes, once a month
Release notes, capacity updates and the occasional deep dive. No fluff, unsubscribe any time.
You're subscribed. Infrastructure notes land in your inbox once a month. Unsubscribe any time.