Infrastructure

The fastest database query is the one you don’t make. Caching is how you turn expensive operations into cheap lookups. Here’s how to do it without shooting yourself in the foot. Cache Placement Where to Cache B r o w F S s a m e ↑ s a r t l e l C s e a t s c t h e → F L a a C s r D ↑ t g N e → A p p l M M i e e c d d a i i t ↑ u u i m m o n C a c h e → D a S L t l a a o r b w g a ↑ e e s s s e t t C a c h e → D a t a b a s e Cache as close to the user as possible, but only what makes sense at each layer. ...

Ansible is deceptively simple. Write some YAML, run it, things happen. Then your playbooks grow, your team grows, and suddenly everything is a mess. Here’s how to write Ansible that scales. Project Structure a ├ ├ │ │ │ │ │ │ │ │ │ ├ │ │ │ ├ │ │ │ └ n ─ ─ ─ ─ ─ s ─ ─ ─ ─ ─ i b a i ├ │ │ │ │ │ └ p ├ ├ └ r ├ ├ └ c └ l n n ─ ─ l ─ ─ ─ o ─ ─ ─ o ─ e s v ─ ─ a ─ ─ ─ l ─ ─ ─ l ─ / i e y e l b n p ├ └ s ├ └ b s w d s c n p e r l t r ─ ─ t ─ ─ o i e a / o g o c e e o o ─ ─ a ─ ─ o t b t m i s t q . r d g k e s a m n t i u c i u h g ├ ├ └ i h g s . e b o x g o i f e c o r ─ ─ ─ n o r y r a n r n r g s t s o ─ ─ ─ g s m v s e s e / i t u / t u l e e s / m o s p a w d s p r s q e n . _ l e a . _ s . l n y v l b t y v . y / t m a . s a m a y m s l r y e b l r m l . s m r a s l y / l v s / m e e l r s s . . y y m m l l Separate inventories per environment. Group vars by function. Roles for reusable logic. ...

Your logs are probably useless. Not because you’re not logging, but because you’re logging wrong. Here’s how to make logs that actually help when things break. The Problem with Unstructured Logs [ [ [ 2 2 2 0 0 0 2 2 2 4 4 4 - - - 0 0 0 3 3 3 - - - 1 1 1 2 2 2 1 1 1 0 0 0 : : : 2 2 2 3 3 3 : : : 4 4 4 5 6 7 ] ] ] I E P N R r F R o O O c : R e : s U s s S i e o n r m g e l t o o h r g i d g n e e g r d w f i e o n n r t u w s r e o r n g 1 2 3 4 5 Try answering these questions: ...

Load balancers are invisible until they break. Then they’re the only thing anyone talks about. Here’s how to get them right. Algorithms That Matter Round Robin Requests go to each server in sequence: A, B, C, A, B, C… 1 2 3 4 5 upstream backend { server 10.0.0.1:8080; server 10.0.0.2:8080; server 10.0.0.3:8080; } Good for: Homogeneous servers, stateless apps Bad for: Servers with different capacities, long-running requests Weighted Round Robin Some servers get more traffic: ...

Everyone has backups. Few people have tested restores. Here’s how to build backup strategies that work when disaster strikes. The 3-2-1 Rule The foundation of backup strategy: 3 copies of your data 2 different storage types 1 copy offsite Example: Production database (primary) Local backup server (different disk) S3 in another region (offsite) This survives disk failure, server failure, and site failure. Database Backups PostgreSQL Logical backups (pg_dump): 1 2 3 4 5 6 7 8 # Full database pg_dump -Fc mydb > backup.dump # Specific tables pg_dump -Fc -t users -t orders mydb > partial.dump # Schema only pg_dump -Fc --schema-only mydb > schema.dump Physical backups (pg_basebackup): ...

You test your application code. Why not your infrastructure? Here’s how to build confidence that your Terraform, Ansible, and Kubernetes configs actually work. Why Test Infrastructure? Infrastructure code has the same problems as application code: Typos break things Logic errors cause outages Refactoring introduces regressions “It works on my machine” applies to terraform too The difference: infrastructure mistakes often cost more. A bad deployment can take down production, corrupt data, or rack up cloud bills. ...

Database migrations are where deployments go to die. One bad migration can corrupt data, lock tables for hours, or bring down production entirely. Here’s how to make them boring. The Golden Rules Every migration must be reversible (or explicitly marked as not) Never run migrations during deploy (separate the concerns) Always test against production-scale data (10 rows works, 10 million doesn’t) Assume the migration will fail halfway (design for it) The Expand-Contract Pattern The safest way to change schemas: expand first, contract later. ...

Your AWS bill is too high. Everyone’s is. The cloud makes it trivially easy to spin up resources and surprisingly hard to know what you’re actually paying for. Here’s how to stop the bleeding. The Low-Hanging Fruit Unused Resources The easiest savings come from things you’re not using. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 # Find unattached EBS volumes aws ec2 describe-volumes \ --filters Name=status,Values=available \ --query 'Volumes[*].[VolumeId,Size,CreateTime]' \ --output table # Find unused Elastic IPs aws ec2 describe-addresses \ --query 'Addresses[?AssociationId==`null`].[PublicIp,AllocationId]' \ --output table # Find idle load balancers (no healthy targets) aws elbv2 describe-target-health \ --query 'TargetHealthDescriptions[?TargetHealth.State!=`healthy`]' Run these monthly. You’ll find forgotten resources every time. ...

You’re a developer. You have smart lights. They turn on when you clap or yell at a cylinder. Congratulations, you’ve automated nothing meaningful. Real home automation is infrastructure. Let’s build it properly. The Stack That Works Home Assistant — The hub. Open source, local-first, integrates with everything. Run it on a Raspberry Pi, a NUC, or a VM. Zigbee/Z-Wave coordinator — Wireless protocols that don’t depend on WiFi or cloud services. Zigbee2MQTT or ZHA for Zigbee, Z-Wave JS for Z-Wave. ...

GitOps sounds simple: Git is the source of truth, automation syncs it to reality. In practice, most teams get it wrong. Here’s how to get it right. The Core Principle GitOps isn’t “we use Git.” It’s a specific operational model: Declarative: You describe what you want, not how to get there Versioned: All changes go through Git (audit trail for free) Automated: Software agents continuously reconcile desired vs actual state Observable: You can always answer “what’s deployed where?” The magic is in reconciliation. Traditional CI/CD pushes changes. GitOps pulls desired state and converges toward it. The system heals itself. ...