Subject
114 entries
Infrastructure
Bookmarks
Ink: infrastructure designed for AI agents
Ink gives AI agents a cloud deployment skill — deploy services, configure domains, manage infrastructure autonomously. One API call to give an agent the ability to ship code to production.
TinyFish: enterprise infrastructure for AI web agents
TinyFish is serverless infrastructure for AI web agents — single API, parallel execution at scale, 98.7% success rate, $0.04/op. Specifically built to handle authenticated websites and dynamic content that breaks standard scraping tools.
Kernel Managed Auth: secure authentication for AI agents
Kernel's Managed Auth gives AI agents a secure way to handle login flows — 2FA, SSO, 1Password integration — without ever exposing credentials to the LLM. Solves one of the least-discussed blockers for autonomous web agents.
OpenClaw Exposure Watchboard: tracking exposed AI agent deployments
The OpenClaw Exposure Watchboard tracks 569K+ publicly reachable AI agent instances — many with leaked credentials, CVEs, and active threat actor interest. Useful for checking if your own deployment is inadvertently public.
Nominal: AI agents for autonomous production incident response
Nominal is AI agents that read your code, watch your infra, and fix production before you wake up — eliminating on-call by 2026. Early access, integrates with AWS/Cloudflare/Vercel/Fly.io and more.
The Self-Driving Codebase: background agents for enterprise delivery
The Self-Driving Codebase thesis: individual developer speed from laptop agents doesn't translate to org velocity. Background agents need cloud isolation, event routing, and governance to actually move DORA metrics.
The Open Anonymity Project: AI user privacy infrastructure
The Open Anonymity Project builds tools and infrastructure for AI user privacy — an open-source effort to give users anonymity options when interacting with AI systems. Still early, but addresses a real gap as AI becomes more central to daily life.
Shuru: local-first microVM sandbox for AI agents
Shuru is a local-first microVM sandbox for AI agents on macOS — gives agents an isolated environment to execute code without touching the host system. Addresses the core risk of agents with shell access running on your actual machine.
Zvec: in-process vector database from Alibaba
Zvec is Alibaba's in-process vector database built on Proxima, their production search engine. No server to run — embed it directly in your app for billion-vector similarity search in milliseconds.
vmux: sandbox runtime for AI agents
vmux is a sandbox runtime for AI agents — preview URLs, log tailing, and tmux attach, running on Cloudflare CPU with Modal GPU backends. Gives agents a managed execution environment rather than direct host access.
Cloudflare Startup Program: $250k Credits
Cloudflare's revamped startup program gives eligible startups up to $250k in credits for Workers, KV, WAF, DDoS protection, and enterprise domains. Founded 2019-2024, $50k-$5M funding, with a bootstrapper path for unfunded companies.
GPU Calculator for LLM Inference
An interactive calculator for estimating GPU requirements for LLM inference — input model size, quantization, batch size, and context length to get memory and throughput estimates. Practical first tool when planning self-hosted or cloud inference deployments.
Internal Constrained PKI: Safe TLS CA for .internal Networks
internal-constrained-pki creates a TLS root CA for .internal networks with X.509 Name Constraints — the CA is cryptographically prevented from signing certificates for public domains, making it safe to share with team members without compromising internet trust.
TensorZero
TensorZero is a self-hosted LLMOps platform that unifies gateway, observability, evaluation, optimization, and A/B testing in one Rust-built system. The feedback loop: production inference data flows back into prompt and model improvement automatically.
Secrets Management Overview
A comparative overview of 10 secrets management tools across on-premises, SaaS, and hybrid deployment models, from open-source options like Ansible Vault and Conjur to commercial platforms like HashiCorp Vault, Doppler, and Akeyless.
Piku: The Tiniest PaaS You've Ever Seen
Piku is a tiny self-hosted PaaS that lets you git-push to deploy to your own servers — a minimal Heroku-style experience with no containers required. Under 2000 lines of Python, runs on any Linux server.
Eidolon: open-source agent service framework
Eidolon is an open-source agent service framework that treats AI agents as first-class services with defined APIs, not just Python scripts — enabling agents to communicate with each other and be composed into larger systems. Targets production deployment of multi-agent architectures.
Open Infrastructure Map: visualizing physical infrastructure from OpenStreetMap
Open Infrastructure Map visualizes power lines, substations, gas pipelines, telecom towers, and other physical infrastructure from OpenStreetMap data. A public, interactive map of the hidden physical systems civilization depends on.
Mazzle: Pipelines as Code for DevOps
Mazzle is a pipelines-as-code tool for building large, complicated, consistent environments — define infrastructure and build pipelines declaratively, run them as a server. An early-generation IaC-adjacent tool predating Terraform's dominance.
Cloud Costs Every Programmer Should Know
Vantage's reference on cloud costs every programmer should internalize — covering compute, storage, networking egress, and database tiers across AWS, GCP, and Azure with rough mental models for estimating costs. Essential calibration for engineers making architectural decisions.
AI's $200B Question
Sequoia Capital's 2023 analysis arguing that GPU capacity was being overbuilt relative to actual AI revenue, with $200B in GPU investment chasing far less in monetizable AI applications. A contrarian note during peak AI hype that proved partially prescient about the infrastructure overhang.
Memgraph Odin
Odin is a Memgraph project — likely a monitoring, schema management, or operational tooling layer for the Memgraph in-memory graph database. Part of Memgraph's ecosystem of tools built around their Cypher-compatible graph DB.
Compute Watch: LLM Compute Costs and GPU Availability Tracker
Compute Watch is a tracker for LLM compute costs and GPU availability — benchmarking inference costs across providers and tracking H100/A100 spot availability. Useful for anyone making infrastructure decisions around model serving costs.
Nvidia H100 GPUs: Supply and Demand
A detailed analysis of the Nvidia H100 GPU supply and demand situation in mid-2023 — how constrained supply was, where the demand was coming from, and what the bottlenecks were. Essential context for understanding the AI infrastructure market that year.
Permit.io: Managed Authorization Service
Permit.io is a managed authorization-as-a-service platform — a hosted alternative to building your own permission system. The pricing page bookmark suggests evaluation for a project requiring fine-grained access control without the operational overhead of self-hosting SpiceDB or OPA.
Monolith: Real Time Recommendation System With Collisionless Embedding Table
Monolith is ByteDance's production recommendation system purpose-built for online training with dynamic sparse features, featuring a collisionless embedding table with expirable embeddings and frequency filtering. It demonstrates that general-purpose ML frameworks like TensorFlow are ill-suited for industrial recommendation with non-stationary data distributions.
The Vector Database Index
Gradient Flow's landscape map of vector databases — published in 2022 when the category was forming, covering Pinecone, Weaviate, Qdrant, Chroma, Milvus, and others. A useful historical snapshot of the vector DB market at the moment it became strategically important.
Banana: Serverless GPU Inference Hosting
Banana is a serverless GPU inference hosting platform for machine learning models — deploy your model in three steps, pay only for inference time. Targets the gap between spinning up a GPU VM yourself and the complexity of cloud ML serving infrastructure.
Alpa: Automated Distributed Training for Large Models
Alpa is a system for automatically parallelizing large neural network training across distributed hardware — finding optimal parallelism strategies without manual configuration. From a Berkeley/CMU research collaboration, it targets the challenge of scaling models beyond single-GPU memory.
YunoHost: Self-Hosted Server for Everyone
YunoHost is an open-source server OS that makes self-hosting web apps as simple as clicking install — Nextcloud, Gitea, Mastodon, and 100+ others available in one catalog. It handles DNS, SSL, user management, and backups automatically, dramatically lowering the barrier to self-hosting.
Hosting a Static Site on Fly.io with Nix and Caddy
A walkthrough of hosting a static site on Fly.io using Nix for reproducible builds and Caddy as the web server — an elegant stack for anyone who wants declarative, reproducible deployments without the complexity of Kubernetes or the cost of AWS.
Banana Serverless Template: Stable Diffusion
Banana.dev's serverless template for deploying Stable Diffusion as an API — package the model, push to Banana, get an API endpoint. One of the first practical paths to serving Stable Diffusion at scale without managing GPU infrastructure.
Point Network: Full Web3 Architecture
Point Network aimed to decentralize all core internet components — domains, storage, and content delivery — unlike web3 projects that kept centralized DNS. It launched mainnet in September 2022 but failed to achieve adoption after the FTX collapse and was acquired by Arweave/AO.
Gazette Core: Unified SQL, Batch, and Stream Processing
Gazette is an open-source platform for building data pipelines that mix SQL, batch, and stream processing in a unified model backed by journals (append-only logs). Designed for organizations that want Kafka-like durability with flexible processing paradigms without committing to a full streaming architecture.
Rancher K3s: Kubernetes on Proxmox Containers
A guide to running K3s (lightweight Kubernetes) inside Proxmox LXC containers — a popular homelab setup that gives you a real Kubernetes cluster on commodity hardware without the overhead of full VMs. Good reference for homelabbers who want production-grade container orchestration locally.
Exploring Production-Ready Recommender Systems with NVIDIA Merlin
NVIDIA Merlin is a framework for building GPU-accelerated production recommender systems — covering feature engineering (NVTabular), training (HugeCTR, Merlin Models), and serving (Triton). This post explores the end-to-end pipeline for large-scale recommendation.
How To F#€k Up An Airport — The BER Podcast
A podcast series documenting the catastrophic mismanagement behind BER — Berlin Brandenburg Airport, which took 14 years to open (vs. the planned 3) and cost 7x its original budget. Required listening for anyone who thinks large public infrastructure projects are boring.
Near-Linear Scaling of Gigantic Model Training on AWS
Amazon Science's post on achieving near-linear scaling when training gigantic language models across thousands of GPUs on AWS infrastructure. Covers parallelism strategies (tensor, pipeline, data) that make training 100B+ parameter models economically feasible.
libaws: AWS Should Be Easy
libaws is Nathan Tippy's opinionated Go CLI that wraps AWS APIs to make them 'easy' — single-command Lambda deployments, simplified IAM, and declarative infrastructure for common patterns. A lightweight alternative to CDK/Terraform for developers who want less abstraction.
Aptos FullNode + Identity Docker Guide
A Docker-based guide to running an Aptos FullNode with identity configuration — practical setup guide for the Aptos testnet period before mainnet launch. Aptos (founded by ex-Meta/Diem engineers) was one of the most anticipated new L1s of 2022.
Warpgate: Smart SSH Bastion
Warpgate is an open-source SSH bastion server written in Rust — sits in front of your SSH targets, handles authentication centrally, and records sessions. Unlike traditional bastion hosts, it works with any standard SSH client without requiring a custom client or VPN.
Running a Bitcoin Lightning Network Node
A guide to running a Bitcoin Lightning Network node — covers the LND or c-lightning stack, channel management, and liquidity. Lightning enables near-instant BTC payments by routing transactions off-chain through a network of payment channels.
DNS Paper (April 2022)
A paper on DNS (Domain Name System) saved in April 2022. The Unix timestamp in the filename (1649775033 = April 12, 2022) suggests it was auto-named at download time. DNS research spans security (cache poisoning, DDoS amplification, DNSSEC), privacy (DNS-over-HTTPS, DNS-over-TLS), and infrastructure reliability.
Kubernetes for MLOps: Scaling Enterprise Machine Learning, Deep Learning, and AI
A book/transcript from the This Week in ML podcast by Sam Charrington on using Kubernetes as the operational backbone for enterprise ML workloads. Covers the full spectrum from containerized training jobs to model serving, making the case that Kubernetes is the de facto standard for scaling ML in production.
Effective Data Science Infrastructure
Ville Tuulos's Manning book on building productive data science infrastructure, with Metaflow as its centerpiece framework. The core argument—that infrastructure exists to make people productive, not to be technically clever—is a useful corrective to the endless tooling churn in ML engineering.
Enclave: Zero-Trust Overlay Networking
Enclave is a zero-trust overlay networking product that creates software-defined private networks between machines without VPN infrastructure — each machine gets a cryptographic identity and point-to-point encrypted tunnels replace network perimeters.
Akash: Decentralized Cloud Computing
Akash Network is a decentralized cloud marketplace where anyone can sell spare compute capacity and anyone can buy it — an open alternative to AWS/GCP with on-chain pricing and Kubernetes-based workload orchestration.
SSH Certificates: Why You're Doing SSH Wrong
Smallstep's case for SSH certificates over authorized_keys — short-lived certificates issued by a CA eliminate the credential sprawl that makes SSH key management a compliance nightmare. The practical alternative to key-based SSH at scale.
EPNS: Decentralized Push Notification Service
EPNS (Ethereum Push Notification Service, now Push Protocol) is a decentralized notification layer for Web3 — enabling dApps to send opt-in push notifications to wallet addresses across chains. Fills the gap between on-chain events and user awareness.
Data50: The World's Top Data Startups
Andreessen Horowitz's 2022 list of the 50 most important data startups — a snapshot of the data infrastructure landscape at peak cloud/ML investment. Useful map of which companies a16z thought were winning in data tooling, observability, and analytics.
Anatomy of an AI System
Kate Crawford and Vladan Joler's 2018 visual essay dissecting an Amazon Echo as a case study in the full global supply chain, labor, and infrastructure behind AI systems — materials, data labor, algorithmic labor, user labor. One of the most cited critical AI artifacts.
How BART on the Golden Gate Bridge Died
An investigation into why BART never crossed the Golden Gate Bridge — exploring the political and engineering decisions that shaped Bay Area transit for decades. A case study in infrastructure decisions and their long-term consequences.
Server Setup 2022
TheOrangeOne's 2022 server setup overview — hardware, OS, containerization, monitoring, backup strategy. A well-documented example of a mature home server stack with emphasis on reliability and observability.
Screw It, I'll Host It Myself
A personal account of moving from cloud services to self-hosted alternatives — the motivations, the setup, and the tradeoffs. Captures the early-2022 wave of interest in digital self-sovereignty driven by concerns about SaaS pricing and data privacy.
The Unbundling of Airflow
fal.ai's analysis of why Apache Airflow is being unbundled by specialized tools — Prefect and Dagster for orchestration, dbt for transformation, Temporal for long-running workflows. The monolithic DAG orchestrator is fracturing along functional lines.
A3MC: Independent Blockchain Research Lab and Validator
A3MC (art3mis.cloud) is an independent blockchain research lab and validator operation — running Solana validators at 7.5% APY with 99.99% uptime, plus participation in Avalanche, The Graph, MultiversX, and Mina Protocol. Enterprise-grade validator infrastructure from a decentralized team.
How to Set Up a Practically Free CDN
A GitHub gist guide to setting up a practically free CDN for personal projects using Cloudflare and free-tier object storage. Covers the stack for serving static assets globally without paying enterprise CDN prices.
Real-Time Machine Learning: Challenges and Solutions
Chip Huyen's definitive survey of real-time ML challenges — the engineering gap between batch ML pipelines and systems that must make predictions on live data with low latency. Covers online prediction, continual learning, and the infrastructure needed to bridge training and serving.
Lessons on ML Platforms from Netflix, DoorDash, Spotify, and More
Ernest Chan's Towards Data Science survey of ML platform design at Netflix, DoorDash, Spotify, and other tech companies — extracting common patterns and lessons from how production ML infrastructure evolved at scale. Practical systems thinking for ML platform builders.
Feature Stores — A Hierarchy of Needs
Eugene Yan's survey of feature stores organized as a hierarchy of needs — from basic feature reuse to online serving to real-time streaming. Practical taxonomy for understanding when you need a feature store and what tier of sophistication your use case requires.
How to Back Up Your Git Repositories
A Medium guide on backing up git repositories — covering bare clone strategy, rsync to external storage, and hosting your own Gitea/Gogs instance. Addresses the risk of losing code when hosting platforms go down or accounts get banned.
Mysten Labs — Sui Blockchain Infrastructure
Mysten Labs is the company building Sui — a new Layer 1 blockchain focused on object-centric state and high-throughput parallel execution. Founded by ex-Meta (Diem/Libra) researchers who spun out after Facebook killed its blockchain project.
Figment Series C — $110M "WAGMI Round"
Figment's $110M Series C announcement (branded as their 'WAGMI Round') — Figment is a leading proof-of-stake staking infrastructure provider supporting validators across 40+ blockchain networks. A signal of institutional capital flowing into PoS validator infrastructure in late 2021.
Inferrd: Managed GPU Model Inference
Inferrd was a managed ML model serving platform with GPU support — deploy models via API without managing infrastructure. A 2021-era MLOps tool in the model deployment category that eventually got acquired.
Ray: A Distributed Framework for Emerging AI Applications
Micah Lerner's paper summary of Ray — UC Berkeley's distributed computing framework for AI workloads. Ray unified task-parallel and actor-based distributed computing in a Python-native API, becoming the foundation for libraries like RLlib, Tune, and later Ray Serve.
The Architecture Behind A One-Person Tech Startup
Anthony Simon's honest account of the infrastructure architecture behind his one-person SaaS business — the actual stack, costs, tradeoffs, and what he'd do differently. Rare for its specificity and honesty about complexity vs. simplicity tradeoffs.
Deterministic Aperture: Twitter's Load Balancing Algorithm
Twitter's Deterministic Aperture load balancing algorithm assigns each client a deterministic subset ('aperture') of backends, reducing connection fan-out while maintaining even load distribution. A principled alternative to round-robin and power-of-two-choices that scales better with horizontal expansion.
Is Your Infrastructure Ready for Learning to Rank?
OpenSource Connections' guide to the infrastructure required before you can deploy learning-to-rank — covering the judgment lists, feature logging pipelines, training data management, and model versioning that most LTR tutorials skip. The operational reality of LTR in production.
Auto Scaling API with AWS Spot Instances
Adapty's guide to designing a scalable API on AWS using spot instances — how to structure your infrastructure to tolerate spot interruptions while dramatically reducing compute costs. A practical architecture pattern for cost-conscious API deployments.
Insight Data Engineering Ecosystem Map
Insight Data Science's map of the data engineering ecosystem circa 2017 — a diagram organizing dozens of tools by pipeline stage (ingest, store, process, visualize). A widely shared snapshot of the Hadoop/Spark era's explosion of competing infrastructure tools.
Google Turns to Machine Learning to Build a Better Data Centre
A 2014 report on Google applying machine learning to optimize data center cooling — one of the first public disclosures that Google was using neural networks to automate infrastructure decisions. DeepMind later published the full methodology in 2016.
Not Just for Search: Using ElasticSearch with Machine Learning Algorithms
An early (2013) case for using Elasticsearch beyond full-text search — specifically as a substrate for machine learning applications like nearest-neighbor lookup and feature indexing. Pre-dates the vector search era but anticipates the same pattern.
Gorgeous, Colorful Look at the Cellphone Signals We Never See
The Verge feature on Nickolay Lamm's colorful visualizations of invisible cell phone signal fields — rendering the electromagnetic environment around cell towers as psychedelic color fields. Makes invisible infrastructure visible.
Data Center Upstarts Take On Hewlett-Packard and EMC
Businessweek on data center upstarts (led by Nutanix) taking market share from HP and EMC in 2013 — the hyperconverged infrastructure wave that packaged compute, storage, and networking into software-defined appliances.
San Francisco's Secret DC Grid
IEEE Spectrum piece on San Francisco's hidden DC power distribution network — a legacy system from the early 20th century still operating underground. An unusual piece of city infrastructure that survived the AC/DC war.
VMware VCDX by the Numbers: Nutanix's Rise
Analysis of VMware VCDX (certification) holders by employer in 2013, showing Nutanix ranking third alongside established data center giants like VCE and EMC. Evidence of Nutanix's rapid enterprise credibility in hyper-converged infrastructure.
The Datacenter as a Computer: Warehouse-Scale Machines
High Scalability's coverage of Google's 'The Datacenter as a Computer' second edition — Barroso and Hölzle's canonical text on warehouse-scale machine design. Defined the vocabulary and engineering tradeoffs for operating entire buildings as programmable compute platforms.
Gartner: AWS Five Times the Size of Cloud Competitors Combined
Gartner's 2013 finding that AWS infrastructure capacity was five times the next 14 cloud providers combined — the market share number that made AWS's dominance narrative concrete and widely quoted. A useful baseline for understanding how lopsided the early cloud market was.
Nutanix OS 3.5: Performance and Style
Nutanix's launch page for NOS 3.5 — the major OS release that added deduplication, compression, and new performance capabilities to their hyper-converged platform. Part of the same August 2013 product push that was challenging EMC and NetApp in enterprise storage accounts.
Introduction to Nutanix Prism UI
Introduction video for Nutanix Prism, the unified management UI for their hyper-converged infrastructure platform — described by the tweet that saved it as 'Beautiful. Fast. Simple.' The UI design was a deliberate differentiator in a market used to complex, fragmented management tools.
Four Things VMware Must Do at VMworld
InformationWeek's 2013 analysis of four things VMware needed to do at VMworld to stay competitive — addressing cloud strategy, OpenStack pressure, hyper-convergence threats, and the transition from a pure hypervisor business to a broader software-defined infrastructure platform.
A Tour Inside CloudFlare's Latest Generation Servers
CloudFlare's 2013 tour of their latest server hardware — SSDs, 10GigE NICs, DRAM cache tiers, and non-RAID storage architecture. Ryan's bookmark note compared it favorably to Nutanix's approach to hyper-converged storage.
Nutanix Engineer Explains Why His Appliance Makes Sense for Analytics
SiliconAngle interview with a Nutanix engineer making the case for why hyper-converged infrastructure is well-suited to analytics workloads in 2013. Nutanix was positioning its appliances as a natural home for Hadoop alongside traditional virtualization.
Implementing Hadoop for Big Data Projects
Inside Analysis piece on practical considerations for implementing Hadoop in enterprise big data projects circa 2013 — covering organizational readiness, hardware choices, and the gap between Hadoop's promise and production reality.
Nutanix Hadoop Solution Brief
Nutanix's Hadoop solution brief describing how NDFS (Nutanix Distributed File System) enables running Hadoop workloads on hyper-converged infrastructure. Marketing collateral saved during Ryan's time at Nutanix.
High Performance at Massive Scale: Lessons Learned at Facebook
Summary of a 2009 Facebook engineering talk on high-performance systems at massive scale — covering their memcached deployment, MySQL sharding, and the operational realities of running at hundreds of millions of users. An early public window into big-company distributed systems practice.
2013: The Year of Converged Infrastructure?
A 2013 enterprise IT blog post declaring it the year of converged infrastructure — pre-packaged stacks combining compute, storage, and networking from a single vendor. A snapshot of when hyper-convergence was becoming mainstream enterprise positioning.
Hadoop on VMware: Another Workload Conquered?
EMC's Chuck Hollis on running Hadoop workloads on VMware virtualization — the argument that bare-metal Hadoop deployments could be replaced by virtualized clusters with manageable performance trade-offs. A 2013 salvo in the bare-metal vs. virtualization debate for big data workloads.
Is There Room for SSDs in the Hadoop Framework?
StorageTuning blog's technical assessment of where SSDs fit in Hadoop's storage hierarchy — evaluating specific bottlenecks (NameNode metadata, shuffle I/O, random reads) where flash storage provides measurable gains over spinning disk.
In 1897, a Bicycle Superhighway Was the Future of California Transit
The 1897 California Cycleway — a proposed elevated wooden bike highway from Pasadena to Los Angeles — as a historical case of cycling infrastructure preceding car dominance. A reminder that the current city form isn't inevitable; cycling was once treated as serious urban transit.
Best Practices for Virtualizing Hadoop
Hadoop Summit presentation on best practices for running Hadoop on VMware HVE (Hadoop Virtualization Extensions) with Hortonworks HDP — addressing the core tension between virtualization flexibility and Hadoop's data locality requirements.
How Intelligent Storage Controllers Have Revolutionized the Industry
DatacenterKnowledge on how intelligent storage controllers — combining flash cache, tiering, and deduplication — changed enterprise storage from dumb arrays into software-defined platforms. A 2013 snapshot of the transition away from spinning-disk-only storage architectures.
Welcome to the Golden Age of Enterprise IT
GigaOm arguing 2013 was the beginning of a golden age for enterprise IT — cloud, SaaS, mobile, and big data converging to make enterprise software interesting again after decades of legacy stagnation. A snapshot of the sentiment that enterprise tech was cool again.
DaaS, MaaS & DRaaS: The Next Phase of Cloud Computing
ReadWrite on the next phase of cloud computing beyond IaaS and SaaS — Desktop-as-a-Service (DaaS), Monitoring-as-a-Service (MaaS), and Disaster Recovery-as-a-Service (DRaaS). A 2013 taxonomy of cloud computing's expanding delivery model landscape.
Benefits Matter — Why I Won't Work for Your YCombinator Start-up
Infrastructure architect Michael Halligan's essay on why experienced engineers won't join YCombinator startups that don't offer real benefits — health insurance, 401k, PTO. A grounded critique of startup compensation culture from an infrastructure practitioner's perspective.
What's Underneath Those Brick Circles?
Burrito Justice's investigation into the brick rings embedded at intersections across San Francisco — remnants of the city's historic streetcar infrastructure. A slice of SF urban archaeology.
Cloudera's Support Team Shares Some Basic Hardware Recommendations
Cloudera's 2010 hardware recommendations for Hadoop clusters — still relevant in 2013 when bookmarked. The canonical guidance on disk, RAM, and CPU specs for commodity Hadoop nodes before cloud deployments became dominant.
In a Nutshell: How OpenStack Works
Victoria Martínez de la Cruz's plain-English explainer of how OpenStack works — breaking down Horizon, Nova, Quantum, Cinder, Glance, Swift, and Keystone into digestible roles. The definitive 2013 quick-reference for the OpenStack component architecture.
Cloud Prizefight: VMware vs. OpenStack
Mirantis's analysis comparing VMware and OpenStack for enterprise private cloud — clearly biased toward OpenStack (Mirantis was an OpenStack services company), but provides a useful breakdown of the real trade-offs between proprietary and open-source virtualization stacks.
Internet Major Outages — January 2013
Personal tweet from January 2013 noting simultaneous major internet service outages affecting Twitter, Amazon, and Facebook — a memorable moment when multiple major platforms went down in close succession.
Cloud Platform Comparison: CloudStack, Eucalyptus, vCloud Director and OpenStack
Network World's 2012 comparison of the four major open-source and commercial cloud platforms — CloudStack, Eucalyptus, vCloud Director, and OpenStack — when enterprises were trying to pick a private cloud standard. The comparison that shows how unsettled the market was before AWS dominance became obvious.
How to Compare VMware and OpenStack
Mirantis SlideShare deck on how to compare VMware and OpenStack for enterprise private cloud decisions — a companion to their widely-read blog post. Useful for the specific evaluation criteria and TCO framing they used to make the OpenStack case.
Seattle SeaFi: Gigabit Fiber Initiative
Seattle Mayor McGinn's 2012 announcement of SeaFi — a plan to bring gigabit fiber to Seattle following Google Fiber's Kansas City deployment. Saved as an example of municipal broadband ambition that followed Google's demonstration of what consumer gigabit could look like.
Small is Beautiful — Nutanix
Nutanix's 2012 blog post arguing that converged, software-defined appliances will displace massive monolithic storage arrays — the 'small is beautiful' thesis against EMC and NetApp incumbents. Retweeted by Vinod Khosla, signaling venture backing for the architectural argument.
The Enterprise: I'm Not Sexy And I Know It
A 2012 TechCrunch piece arguing that enterprise infrastructure startups are unfairly dismissed as unsexy — the author points to Nutanix's distributed hypervisor approach as an example of genuine technical innovation happening outside consumer tech.
Facebook Seeks Next-Generation Big Data Tools
InformationWeek's 2012 report on Facebook's efforts to move beyond Hadoop for its next-generation big data infrastructure — one of the first major signals that the tech industry's heaviest Hadoop user was already looking past it.
Data Centers Waste Vast Amounts of Energy
NYT investigation into the massive energy waste inside data centers — servers running at 6-12% utilization, backup systems idling on standby, cooling systems sized for peak loads that rarely arrive. Exposed the gap between the clean, efficient image of cloud computing and the physical reality.
Mellanox Introduces RDMA-Based Virtual SAN Software Appliance
Mellanox announced a Virtual SAN software appliance built on RDMA that claimed 6x the performance of traditional Fibre Channel SAN hardware. A signal that software-defined storage was coming for legacy enterprise SAN.
Amazon Glacier: Ultra-Low-Cost Archival Storage
Amazon Glacier launched in 2012 as an ultra-low-cost archival storage service at $0.01/GB/month — an order of magnitude cheaper than S3. The tradeoff was retrieval latency of hours, not milliseconds, making it purely for cold data.
Not Just for a Rainy Day: The True Meaning of Converged
Nutanix's 2012 blog post making the case for converged infrastructure — combining compute, storage, and networking in a single appliance rather than maintaining separate silos. Written during their early go-to-market phase, when hyperconvergence was still a novel concept rather than an industry standard.
A Rough Guide to Keeping Your Website Up Through High Traffic
Rainforest's 2012 guide to keeping a web application running through traffic spikes — covering CDN, caching, database connection pooling, and graceful degradation. Early practical devops writing from the era before managed auto-scaling became trivial.
The Lowline: New York's Underground Park of the Future
The Lowline project proposed converting an abandoned 1908 trolley terminal on Manhattan's Lower East Side into an underground park using fiber-optic sunlight delivery technology. An inverse of the High Line — bringing parkland below street level into existing dead infrastructure.
Deploying ElasticSearch with Chef Solo
An early 2012 tutorial for deploying Elasticsearch using Chef Solo — a document from when both Elasticsearch and infrastructure-as-code were early in their mainstream adoption curves. Shows the 2012 devops toolchain before Ansible and Terraform emerged.
How the FRA Is Regulating Passenger Rail Out of Existence
The East Bay Bicycle Coalition's analysis of how the FRA's freight-rail safety standards, when applied to passenger rail, mandate impossibly heavy trains that can't share infrastructure with light vehicles — explaining why US passenger rail is so expensive, slow, and rare compared to Europe and Japan. A case study in regulatory capture killing an industry.