# infra Published articles for infra. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories DevFeed: [AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories]() Original publisher: [Read original article]() Author: NVIDIA Writers Published: 2026-09-15T16:55:40Z Content type: news Language: en Sources: [NVIDIA Blog]() Topics: [AI Infrastructure](), [Nvidia](), [DSX](), [Vera Rubin](), [Low-Latency Inference](), [NVLink](), [Conversational AI]() Tags: [ai](), [ai-infrastructure](), [dsx](), [inference](), [infra](), [latency](), [low-latency-inference](), [networking](), [nvidia](), [nvidia-blackwell](), [nvidia-dsx](), [nvidia-vera](), [nvidia-vera-rubin](), [nvlink]() ### AI overview NVIDIA's AI Infra Summit coverage describes collaborations and platform updates focused on improving AI factory efficiency. The article highlights Vera Rubin systems, DSX MaxLPS, Dynamo inference software, NVLink and networking technologies, including claims of up to 1.4x more tokens per megawatt through factory-wide power optimization. ### Source excerpt Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA, Tuesday spoke on AI factory efficiency at the AI Infra Summit, the Santa Clara Convention Center event that has morphed into a Coachella of infrastructure tech. Before a packed audience -- with more than 8,000 attendees this year, up from 3,500 last year -- [...] ## Teravolt looks to cannibalize older industries to meet AI power demand DevFeed: [Teravolt looks to cannibalize older industries to meet AI power demand]() Original publisher: [Read original article]() Author: Thomas Claburn Published: 2026-09-14T13:45:00Z Content type: article Language: en Sources: [www.theregister.com - Articles]() Topics: [Artificial Intelligence](), [datacenter](), [Bitcoin]() Tags: [ai](), [ai-and-ml](), [artificial-intelligence](), [bitcoin](), [datacenter](), [electricity-grid](), [energy](), [infra]() ### AI overview Teravolt is considering repurposing Bitcoin farms, aluminum smelters, and other older infrastructure as datacenters to help meet growing AI power demand. ### Source excerpt Bitcoin farms, aluminum smelters, and other old infra is more lucrative to repurpose as a datacenter ## Training 100x Cheaper Retrieval models Neon and Castform DevFeed: [Training 100x Cheaper Retrieval models Neon and Castform]() Original publisher: [Read original article]() Author: Pranav Aurora Published: 2026-08-05T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Retrieval Augmented Generation (RAG)](), [Training AI Models](), [Machine learning](), [Open Source Models & Datasets](), [Large Language Model](), [Open Source](), [Latency](), [Database]() Tags: [agentic](), [agents](), [cost](), [embedding](), [infra](), [llms](), [machine-learning](), [model](), [models](), [open-source](), [performance](), [post-training](), [product](), [rag](), [retrieval](), [rl](), [scale](), [search](), [systems](), [tools](), [training]() ### AI overview The article explains how Castform uses reinforcement-learning post-training to improve open-weight models for agentic retrieval. It contrasts multi-step retrieval with one-shot embedding search, emphasizing the cost and latency of repeated frontier-model calls and the potential for smaller open models to perform specific search tasks more cheaply. ### Source excerpt "Most teams' best training data is just sitting in their databases. The problem is that turning raw data into something usable is hard, and letting agents read, search, and mutate data cheaply at scale requires advanced infra. Pointing Castform at Neon skips both." -- Ying Hang Seah, cofounder, Castform ## Why platform engineers observe infra and help devs monitor apps DevFeed: [Why platform engineers observe infra and help devs monitor apps]() Original publisher: [Read original article]() Author: Sam Barlien Published: 2026-07-23T05:40:01Z Content type: article Language: en Sources: [Platform Engineering Blog]() Topics: [Platform Engineering](), [Monitoring](), [OpenTelemetry](), [distributed-systems](), [Kubernetes](), [CI/CD]() Tags: [ci-cd](), [infra](), [infrastructure](), [kubernetes](), [monitoring](), [opentelemetry](), [platform](), [platform-engineering]() ### AI overview This article explains the dual mandate of platform engineering: maintaining visibility into shared infrastructure while enabling developers to monitor their own applications. It argues that infrastructure-only monitoring misses application-level failures in unpredictable self-service environments and presents OpenTelemetry as an enabler of monitoring at scale. ### Source excerpt Learn why platform engineers must observe infrastructure and empower developers with effortless app monitoring and how OpenTelemetry enables both at scale. ## Pale Blue Spring Admin adds CRUD editing and custom UI snippets DevFeed: [Pale Blue Spring Admin adds CRUD editing and custom UI snippets]() Original publisher: [Read original article]() Author: Mike Yerou Published: 2026-05-14T11:57:06Z Content type: release Language: en Sources: [Rock and Null]() Topics: [CRUD](), [Spring Boot](), [Django Admin](), [jpa](), [Kotlin](), [ui]() Tags: [backend](), [crud](), [django-admin](), [infra](), [jpa](), [kotlin](), [spring-boot](), [ui]() ### AI overview The latest Pale Blue Spring Admin release adds create, update, and delete operations directly in the UI. It also introduces custom UI snippets for adding elements such as dashboard widgets and action buttons, extending the admin portal beyond its generated interface. ### Source excerpt Read-only is great, but editing is better. The latest update to Pale Blue Spring Admin introduces full CRUD functionality and custom UI snippets, making your Spring Boot admin panel as powerful as your favorite Python tools. ## How KIKO Milano scales for Black Friday DevFeed: [How KIKO Milano scales for Black Friday]() Original publisher: [Read original article]() Author: Meghan Schaefer Published: 2026-05-05T07:00:00Z Content type: article Language: en Sources: [Vercel News]() Topics: [Vercel](), [Amazon EC2](), [Amazon Web Services](), [build times](), [Provisioning](), [configuration](), [App](), [releases]() Tags: [aws](), [configuration](), [infra](), [infrastructure](), [performance](), [provisioning](), [releases](), [vercel]() ### AI overview KIKO Milano moved its ecommerce application from manually scaled AWS infrastructure to Vercel. The migration eliminated three weeks of Black Friday infrastructure preparation, reduced application build times by 75%, and enabled the team to deploy multiple times per day. Automatic scaling and on-demand environments allowed the team to focus on campaigns, testing, and the ecommerce experience instead of infrastructure limits. ### Source excerpt KIKO Milano on Vercel: Eliminated 3 weeks of Black Friday infrastructure prep 75% decrease in app build times Went from minimal releases to deploying multiple times per day KIKO Milano's ecommerce team used to treat peak traffic as an operations project. Weeks before Black Friday, they had to manually scale AWS infrastructure and adjust application configuration, knowing that if demand exceeded forecasts, their site could slow down or even break, costing them real revenue. After migrating to Vercel, they removed manual prep from their playbook. Instead of provisioning infrastructure, their team now ships campaigns and tests daily. The Black Friday problem: manual scaling and stressful failure modes Before Vercel, KIKO ran their ecommerce app on AWS EC2 instances sized for normal traffic, then scaled by hand ahead of peak periods. This was their previous process leading into Black Friday: Manual scaling window: Black Friday prep began 2-3 weeks ahead of the event, then had to be unwound after the traffic spike passed. Infrastructure configuration: The infra team manually adjusted AWS EC2 capacity to support expected traffic. Application configuration: Scaling also required application-side changes, making peak readiness more than just an infrastructure task. War room prep: Chat channels and conference rooms were set up for teams to triage performance and downtime problems. Beyond the manual prep work, KIKO's dev team had to anticipate failure modes that hurt the business. Slow navigation and site instability made it hard for users to purchase, and, in Ant's words, they knew that at any point, "everything could break." Worry-free delivery and automatic scaling Vercel changed the work from preparation to execution. With environments available on demand, faster builds, and automatic scaling during traffic spikes, Ant's team stopped planning around infrastructure limits and focused on the ecommerce experience itself. Environments without the bottleneck KIKO's old setup r ## Wix Migrated Its MySQL EC2 Fleet from Intel CPUs to Graviton DevFeed: [Wix Migrated Its MySQL EC2 Fleet from Intel CPUs to Graviton]() Original publisher: [Read original article]() Author: Wix Engineering Published: 2026-03-23T11:27:39Z Content type: article Language: en Sources: [Wix Engineering]() Topics: [Graviton](), [MySQL](), [Ansible](), [awx](), [Amazon EC2](), [intel](), [Databases](), [DevOps](), [Replication](), [Server](), [Linux]() Tags: [amazon](), [ansible](), [awx](), [devops](), [ec2](), [graviton](), [infra](), [intel](), [linux](), [mysql](), [rdbms](), [replication]() ### AI overview Wix's DB Infra team migrated more than 1,000 MySQL EC2 servers across over 160 clusters from Intel-based CPUs to Graviton in one month, ahead of the original three-month timeline. The project also enabled a move to Amazon Linux 2023 and away from CentOS 7.0, using AWX workflows and Wix's Dev Portal to automate the process while accounting for replication and cluster reliability concerns. ### Source excerpt Last January, the DB Infra team at Wix embarked on a strategic and very important mission: Migrating all of our MySQL EC2 servers from Intel based CPUs to the new shiny Graviton ones. Not only will this allow us to work on better CPUs, saving us money and improving our databases, we'll be able to move to Amazon Linux 2023 OS and move from the EOL CentOS 7.0. That sounds simple enough, until you realize that we're dealing with over 1K servers spread over 160+ MySQL clusters. The initial... ## How Merkle Science fights crypto crime at scale with ClickHouse Cloud DevFeed: [How Merkle Science fights crypto crime at scale with ClickHouse Cloud]() Original publisher: [Read original article]() Author: Akshay Gupta, Lead Data Engineer, and Priyanshu Sehgal, Sr. Data Engineer Published: 2025-11-13T09:30:52Z Content type: article Language: en Sources: [ClickHouse Blog]() Topics: [clickhouse](), [Cloud](), [Data Infrastructure](), [Blockchain](), [threat detection](), [real-time](), [on-prem]() Tags: [blockchain](), [clickhouse](), [cloud](), [data-infrastructure](), [infra](), [on-prem](), [real-time](), [threat-detection]() ### AI overview Merkle Science uses ClickHouse as a core part of its blockchain analytics and predictive risk platform for detecting, tracing, and preventing illicit activity. The company ingests about 1.5 terabytes of data weekly and runs more than 18 million queries, relying on ClickHouse for transaction monitoring, wallet attribution, cross-chain forensics, and real-time threat detection. The article describes its migration from self-managed ClickHouse infrastructure to ClickHouse Cloud to simplify operations and support faster scaling. ### Source excerpt "Even if data increases by 4x or 5x in the next year, we literally don't have to worry about it. We've got the architecture right, the stack right, and the right platform at the end of the day." - Akshay Gupta, Lead Data Engineer ## Branching as the New Standard for Relational Databases DevFeed: [Branching as the New Standard for Relational Databases]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2025-07-10T16:51:48Z Content type: opinion Language: en Sources: [Blog -- Neon Docs]() Topics: [Databases](), [Development](), [AI Development](), [cloud-infrastructure]() Tags: [agent](), [ai](), [databases](), [development](), [infra](), [migrations](), [product](), [relational-databases](), [software-development](), [workflows]() ### AI overview The article argues that relational databases have lagged behind the software stack in iteration speed, automation, and developer experience. It presents database branching as a new standard for development workflows, especially because AI agents need databases that are fast to provision, scalable, integrated, cost-efficient, and easy to discard. ### Source excerpt Over the last decade, nearly every part of the software development stack has evolved to support faster iteration, better automation, and less oversight. But one layer has stubbornly resisted this evolution - the relational database. The stack evolved, the database stayed behind... ## How Hugging Face Scaled Secrets Management for AI Infrastructure DevFeed: [How Hugging Face Scaled Secrets Management for AI Infrastructure]() Original publisher: [Read original article]() Author: Thomas Segura Published: 2025-03-31T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [migration](), [iac-security](), [incident]() Tags: [ai-infrastructure](), [automation](), [aws](), [azure](), [case-study](), [cloud](), [gcp](), [hugging-face](), [incident](), [infra](), [infrastructure-as-code](), [kubernetes](), [migration](), [partnerships](), [secrets](), [security](), [terraform]() ### AI overview A case study of Hugging Face migrating secrets management to Infisical for a multi-cloud infrastructure. It covers secret sprawl, RBAC and SSO needs, local-development workflows, secret rotation, Terraform, and Kubernetes secret updates. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## How Zap.xyz Built a Serverless CDC Pipeline with Neon and Inngest DevFeed: [How Zap.xyz Built a Serverless CDC Pipeline with Neon and Inngest]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2025-02-12T13:39:09Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [data]() Tags: [amazon-aurora](), [amazon-rds](), [architecture](), [aws](), [case-studies](), [data](), [database](), [developer-experience](), [events](), [infra](), [postgres](), [real-time](), [scale](), [serverless](), [streams](), [vercel](), [workflows]() ### AI overview Zap.xyz describes replacing an AWS-based change data capture approach with Neon and Inngest for a serverless pipeline that processes incoming community-channel messages and triggers follow-up workflows. ### Source excerpt "Inngest + Neon handle the entire change data capture process for us. Setting it up took a fraction of the time compared to AWS" (Jacob Devore, Co-Founder at Zap) Zap.xyz is a new crypto aggregation platform that processes millions of messages from community channels like Telegra... ## Efficient Dockerfile templating for complex build scenarios DevFeed: [Efficient Dockerfile templating for complex build scenarios]() Original publisher: [Read original article]() Author: Tom Published: 2025-01-01T00:00:00Z Content type: tutorial Language: en Sources: [Tomasz Gągor]() Topics: [Dockerfile](), [Docker](), [Docker Image](), [configuration]() Tags: [alpine](), [base-images](), [best-practices](), [build](), [dependencies](), [devops](), [docker](), [docker-base-images](), [dockerfiles](), [go](), [infra](), [java](), [linux](), [scm](), [sre](), [ubuntu]() ### AI overview This article explains why Dockerfile templating can help maintain complex Docker base-image variants. It discusses combinations of operating systems, Tomcat versions, Java versions, configurations, and package names, along with the differing responsibilities of development and infrastructure teams. ### Source excerpt Why even consider templating Dockerfiles? Dockerfiles revolutionized the industry with their simplicity. Each instruction creates a new layer in the image, which is automatically cached. This process integrates well with SCM, where you "commit" the results of one stage and move forward with other changes. The process can be easily parameterized with ARG instructions, similar to ENV but provided during the build. This allows for creating highly flexible builds. For most users, this is more than sufficient. However, there's a notable exception: Docker base images. ## Building On-call: Our observability strategy DevFeed: [Building On-call: Our observability strategy]() Original publisher: [Read original article]() Author: Martha Lambert Published: 2024-08-22T15:33:00Z Content type: article Language: en Sources: [The incident.io Blog]() Topics: [observability](), [incident](), [Availability](), [User experience (UX)](), [dashboards](), [log management]() Tags: [availability](), [building](), [dashboards](), [engineering](), [health](), [high-availability](), [incident](), [incident-channel](), [incident-management](), [incident-response](), [infra](), [lens](), [logs](), [observability](), [on-call](), [outage](), [post](), [post-mortem](), [reactive](), [slack-incident](), [ux]() ### AI overview incident.io explains how observability supports its on-call product by helping engineers proactively detect degrading systems and react quickly when incidents occur. The article emphasizes that effective observability depends not only on collecting data, dashboards, and logs, but also on making the information easy for engineers to find and use. It presents a user-focused, product-like approach intended to improve adoption and make observability a shared engineering responsibility. ### Source excerpt Our customers count on us to sound the alarm when their systems go sideways--so keeping our on-call service up and running isn't just important; it's non-negotiable. To nail the reliability our customers need, we lean on some serious observability (or as the cool kids say, o11y) to keep things running smoothly. ## Making Machines Move DevFeed: [Making Machines Move]() Original publisher: [Read original article]() Published: 2024-07-30T00:00:00Z Content type: article Language: en Sources: [The Fly Blog]() Topics: [migration]() Tags: [cdn](), [close-to-users](), [cloud](), [deploy-app-servers](), [docker](), [elixir](), [fly](), [fly-io](), [heroku-alternative](), [heroku-competitor](), [hosting](), [i](), [infra](), [migration](), [networking](), [postgresql-clusters](), [servers](), [storage]() ### AI overview Fly.io describes the challenge of migrating stateful virtual machines whose attached NVMe volumes anchor application data to a physical worker. The article contrasts simple stateless-worker draining with volume migration, where backup restoration risks data loss and copying volumes can cause unacceptable interruption. ### Source excerpt We're Fly.io, a global public cloud with simple, developer-friendly ergonomics. If you've got a working Docker image, we'll transmogrify it into a Fly Machine: a VM running on our hardware anywhere in the world. Try it out; you'll be deployed in just minutes. At the heart of our platform is a systems design tradeoff about durable storage for applications. When we added storage three years ago, to support stateful apps, we built it on attached NVMe drives. A benefit: a Fly App accessing a file on a Fly Volume is never more than a bus hop away from the data. A cost: a Fly App with an attached Volume is anchored to a particular worker physical. bird: a BGP4 route server. Before offering attached storage, our on-call runbook was almost as simple as "de-bird that edge server", "tell Nomad to drain that worker", and "go back to sleep". NVMe cost us that drain operation, which terribly complicated the lives of our infra team. We've spent the last year getting "drain" back. It's one of the biggest engineering lifts we've made, and if you didn't notice, we lifted it cleanly. The Goalposts With stateless apps, draining a worker is easy. For each app instance running on the victim server, start a new instance elsewhere. Confirm it's healthy, then kill the old one. Rinse, repeat. At our 2020 scale, we could drain a fully loaded worker in just a handful of minutes. You can see why this process won't work for apps with attached volumes. Sure, create a new volume elsewhere on the fleet, and boot up a new Fly Machine attached to it. But the new volume is empty. The data's still stuck on the original worker. We asked, and customers were not OK with this kind of migration. Of course, we back Volumes snapshots up (at an interval) to off-network storage. But for "drain", restoring backups isn't nearly good enough. No matter the backup interval, a "restore from backup migration" will lose data, and a "backup and restore" migration incurs untenable downtime. The next thought you have is, ## How to reduce Postgres compute costs as you scale DevFeed: [How to reduce Postgres compute costs as you scale]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2024-04-29T15:31:21Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Amazon RDS](), [Databases](), [Amazon Web Services](), [Availability](), [autoscaling]() Tags: [amazon-rds](), [article](), [autoscaling](), [availability](), [aws](), [cloud](), [cost](), [high-availability](), [how-to](), [infra](), [performance](), [postgres](), [pricing]() ### AI overview The article explains how to reduce Postgres compute costs as database demand grows. It focuses primarily on Amazon RDS, describing how instance size, on-demand versus reserved pricing, and single-AZ versus multi-AZ deployments affect compute bills. It also presents Neon's scale-to-zero and autoscaling capabilities as ways to pay for the compute resources actually used. ### Source excerpt Reducing cloud costs is on everyone's mind. While storage prices continue to drop, fewer innovations have been made to reduce compute costs, even though compute typically consumes a larger chunk of the bill. That's why we've built Neon with scale to zero and autoscaling, so that...