# S3 Published articles for S3. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## STRG Review: Can It Make Cloud Storage Costs More Predictable? DevFeed: [STRG Review: Can It Make Cloud Storage Costs More Predictable?]() Original publisher: [Read original article]() Author: WDD Staff Published: 2026-11-06T13:46:00Z Content type: article Language: en Sources: [Web Designer Depot]() Topics: [Cloud](), [Amazon S3](), [MSP MCP](), [cost-optimization](), [restic]() Tags: [aes](), [agents](), [ai](), [ai-agents](), [ai-tools](), [ai-workloads](), [api](), [aws-cli](), [aws-s3-alternatives](), [backups](), [budgeting](), [candidate](), [cloud](), [cloud-infrastructure](), [cloud-storage](), [compatibility](), [connect](), [console](), [costs](), [developer-tools](), [egress](), [mcp](), [object-storage](), [permissions](), [s3](), [s3-compatible-storage](), [sponsored](), [sponsored-review](), [strg]() ### AI overview STRG's main cost advantage is a published $6.80 per TB monthly rate that includes egress and standard S3 requests, with the first 10 GB free and no retrieval fees or minimum storage duration listed. It supports familiar S3 tools and common object operations, but features such as SSE-KMS, IAM roles, tagging, and static website configuration are not listed as supported. Its console Assistant and MCP Server let customers scope agent access through selected credentials and permissions; STRG says prompts and customer data are not used to train general-purpose models. STRG also describes encryption, integrity checks, Object Lock, and optional cross-region replication, with EU-Central and US-East storage regions. The excerpt recommends estimating actual workload costs and piloting required operations, resilience, and recovery. The published price excludes taxes, larger deployments may have separate terms, and standard availability commitments or service credits require an executed agreement. ### Source excerpt Cloud storage has a catch: uploading your files is only part of the bill. STRG says it bundles storage, egress, and standard S3 requests for $6.80 per TB per month--and lets AI agents manage your storage with permissions you control. ## Oracle to S3 as Parquet: Export Tables and Replicate Changes with CDC DevFeed: [Oracle to S3 as Parquet: Export Tables and Replicate Changes with CDC]() Original publisher: [Read original article]() Author: Seán Whelan Published: 2026-11-06T12:27:12Z Content type: article Language: en Sources: [Estuary]() Topics: [streaming data to S3 with Parquet](), [parquet](), [Parquet files in Amazon S3](), [Oracle Database](), [AWS Glue](), [Debezium Connector]() Tags: [aws](), [aws-glue](), [cdc](), [changes](), [compare](), [dms](), [export](), [goldengate](), [managed](), [oracle](), [parquet](), [pipeline](), [replicate](), [s3](), [tables]() ### AI overview Oracle-to-S3 Parquet choices depend on whether you need change capture: DMS and log-based CDC tools read redo and capture deletes, while exports and scheduled query jobs see surviving rows only. All approaches must handle Oracle type mapping and immutable Parquet files, where updates and deletes become new rows; current-state queries therefore deduplicate by key and filter the latest delete. The excerpt details DMS setup, including archived redo and supplemental logging, Parquet and timestamp settings, bounded LOB sizing, and file batching. It also describes Estuary's LogMiner capture and S3 materialization: backfill runs alongside redo mining, changes append on a configurable upload interval, and deletes and key changes are represented as rows. Estuary's stated tradeoffs include skipped LOB and several other column types, string or numeric mappings that may need configuration, no hard-delete or truncate handling, and retained historical files. Alternatives include Autonomous Database exports for simple full-query exports, and Spark or Glue for custom transforms; those query-based approaches do not capture deletes. ### Source excerpt Move Oracle to S3 as Parquet with AWS DMS, AWS Glue, GoldenGate or a managed CDC pipeline. Compare how each handles NUMBER, DATE, LOBs and deletes. ## S3 to Kafka: How to Turn New S3 Files into Kafka Messages DevFeed: [S3 to Kafka: How to Turn New S3 Files into Kafka Messages]() Original publisher: [Read original article]() Author: Seán Whelan Published: 2026-11-06T12:27:02Z Content type: article Language: en Sources: [Estuary]() Topics: [Apache Kafka](), [Confluent Cloud](), [Amazon MSK](), [parquet]() Tags: [aws](), [aws-lambda](), [compare](), [files](), [how-to](), [kafka](), [managed](), [messages](), [ordering](), [pipeline](), [s3](), [setup](), [source](), [spark]() ### AI overview S3 files can become Kafka messages through Kafka Connect, S3 event notifications with Lambda, Spark or Flink, or a managed pipeline. The main differences are how each discovers and tracks files, supported parsing and transformation, and setup; most approaches deliver at least once, so consumers may need to handle duplicates. Listing tools can miss late-arriving keys when they track only the last key, while re-reading modified files can produce duplicates. The guide details S3 and Kafka permissions, format and message-size constraints, and operational tradeoffs. Estuary's described route lists a bucket prefix, parses new files into a durable collection, then materializes records to Kafka; it runs every five minutes by default, delivers at least once, and has specific Parquet size, field truncation, schema-subject, authentication, and deletion limitations. Event-triggered Lambda is positioned for a modest number of uploads needing quick reaction, and Spark or Flink for files requiring transformation. ### Source excerpt Move S3 data into Kafka with an S3 source connector, AWS Lambda, Spark or a managed pipeline. Compare file ordering, formats, duplicates and setup. ## Salesforce to S3: Export Objects as Parquet and Capture Changes DevFeed: [Salesforce to S3: Export Objects as Parquet and Capture Changes]() Original publisher: [Read original article]() Author: Seán Whelan Published: 2026-11-06T12:26:46Z Content type: tutorial Language: en Sources: [Estuary]() Topics: [Parquet files in Amazon S3](), [parquet](), [Amazon S3](), [AWS Glue]() Tags: [17](), [2](), [3](), [about](), [amazon](), [aws](), [aws-glue](), [capture](), [changes](), [compare](), [export](), [fields](), [managed](), [parquet](), [pipeline](), [python](), [s3](), [salesforce](), [types](), [with]() ### AI overview Salesforce-to-S3 Parquet exports rely on an initial object load followed by polling `SystemModstamp` or consuming change events; updates append new rows, and deletes appear only when the method includes deleted records. The choice depends on scale and destination: AppFlow suits a few objects in AWS but has a 15 GB per-run cap and needs `preserveSourceDataTyping` enabled to retain types; Glue zero-ETL writes Iceberg tables that can be compacted but omits some objects; Python allows custom logic; managed pipelines suit many objects or destinations. Plan for shared Salesforce API limits, per-object S3 prefixes, and incremental recovery. Formula values can change without `SystemModstamp`, so they need periodic rereads; append-only files then require a query that combines the latest full row with refreshed formula values. Purged records cannot be recovered, and reloading can duplicate rows. Validate types, edits, deletes, and new fields after loading. ### Source excerpt Move Salesforce to S3 as Parquet with Amazon AppFlow, AWS Glue, Python or a managed pipeline. Compare how each handles types, formula fields and deletes. ## Dropbox ditched AWS S3 for its own in-house system. Now it's buying Tigris to sell S3-compatible storage to AI developers. DevFeed: [Dropbox ditched AWS S3 for its own in-house system. Now it's buying Tigris to sell S3-compatible storage to AI developers.]() Original publisher: [Read original article]() Author: Paul Sawers Published: 2026-11-05T14:00:00Z Content type: news Language: en Sources: [The New Stack]() Topics: [dropbox](), [Magic Pocket](), [Amazon S3]() Tags: [acquisition](), [ai](), [ai-applications](), [ai-developers](), [ai-infrastructure](), [ai-models](), [ai-powered](), [aws](), [aws-s3](), [chatgpt](), [claude](), [cloud](), [cloud-services](), [cloud-storage](), [compute](), [developers](), [dropbox](), [infrastructure](), [s3](), [startup](), [storage]() ### AI overview Dropbox plans to acquire Tigris for $350 million and use its S3-compatible object-storage technology as the foundation for DBX Agent Cloud, an upcoming service aimed at developers building AI applications. Dropbox says operating its own storage for its users differs from selling storage to external customers; Tigris brings a commercial product, customers, and developer relationships. Tigris keeps data close to compute across cloud and compute environments and charges no egress fees for transfers out of its service, addressing costs and delays that can arise when AI workloads run outside AWS using Amazon S3. Dropbox plans to add filesystem and content-processing capabilities after launching standalone object storage. The acquisition awaits regulatory approvals, and Dropbox has not announced a launch date or detailed the technical integration. Existing Tigris customers are promised continued support without an initial migration requirement; later changes remain uncertain. ### Source excerpt Dropbox is entering the AI infrastructure space in a big way, announcing plans on Thursday to acquire Tigris, a startup The post Dropbox ditched AWS S3 for its own in-house system. Now it's buying Tigris to sell S3-compatible storage to AI developers. appeared first on The New Stack. ## MongoDB to S3: Parquet Files, Change Streams & Nested Documents Guide DevFeed: [MongoDB to S3: Parquet Files, Change Streams & Nested Documents Guide]() Original publisher: [Read original article]() Author: Seán Whelan Published: 2026-11-04T22:41:24Z Content type: article Language: en Sources: [Estuary]() Topics: [streaming data to S3 with Parquet](), [MongoDB](), [Parquet files in Amazon S3](), [Amazon S3](), [parquet](), [AWS Database Migration Service](), [AWS Glue](), [compressing data with Parquet](), [DLT](), [Amazon DocumentDB]() Tags: [change](), [mongodb](), [nested](), [parquet](), [parquet-files](), [s3]() ### AI overview MongoDB change streams can feed append-only Parquet files in S3, but updates and deletes become new rows; readers must retain the latest row per `_id` and filter delete markers to represent current state. Change streams require a replica set or sharded cluster and an oplog window long enough to cover outages. A time-window copy or batch job misses deletes. Atlas Data Federation, AWS DMS, custom Spark/Glue/dlt jobs, and managed CDC services differ in setup and how they map nested BSON data; structs, flattened columns, and JSON text are the described options. Schema changes affect later files, and Decimal128 or digit-only strings can lose precision or leading zeros unless written as suitable decimal or string types. Estuary's described route captures changes durably and appends Parquet on a configurable interval, with nested-field depth settings; its Decimal128 and integer inference behavior may require `castToString`. The excerpt also gives S3 permissions, lifecycle, validation, and recovery requirements, including re-reading collections after an oplog history gap. ### Source excerpt Move MongoDB to S3 as Parquet with Atlas Data Federation, AWS DMS, Spark or a managed pipeline. Compare how each handles nested documents and deletes. ## wasmCloud 2.12: A workload on every host, secrets a plugin serves, and a pool sized to the machine DevFeed: [wasmCloud 2.12: A workload on every host, secrets a plugin serves, and a pool sized to the machine]() Original publisher: [Read original article]() Author: Eric Gregory Published: 2026-11-04T00:00:00Z Content type: release Language: en Sources: [wasmCloud]() Topics: [wasmCloud](), [Secrets Management](), [memory pool](), [upgrade](), [secret rotation](), [control-plane](), [CI/CD](), [HashiCorp Vault](), [infrastructure monitoring]() Tags: [32-bit](), [403](), [authentication](), [backend](), [changes](), [client](), [crd](), [machine](), [node](), [plugin](), [plugins](), [pool](), [release](), [s3](), [scale](), [secrets](), [serve](), [sigstore](), [tags](), [wasmcloud]() ### AI overview wasmCloud 2.12 adds Kubernetes workload and secret-management capabilities alongside runtime and security changes. `WorkloadDaemonSet` maintains the configured number of instances on every matching host, while `/scale` lets operators scale a standalone `WorkloadReplicaSet`; a deployment-owned replica set should be scaled through its `WorkloadDeployment`. Host plugins can serve secrets to other plugins, with grants scoped by workload namespace and name. Rotated backend values reach workloads at their next bind without a volume remount, while adding keys, changing grants, or changing backend access requires a host rollout. Before upgrading the chart, apply all 2.12.0 CRDs because Helm upgrade does not update them. Other notable changes include a machine-sized Wasmtime pool, optional guest-memory growth accounting, S3 blobstore support, NATS authentication and safer TLS behavior, and signed release tags. The release includes breaking changes for CRDs, NATS settings, Redis binding configuration, and embedder APIs. ### Source excerpt wasmCloud 2.12 adds the WorkloadDaemonSet CRD and a /scale subresource on WorkloadReplicaSet, lets a host plugin serve secrets to other plugins, sizes the instance pool from the node, adds an S3 blobstore backend and NATS client authentication, and signs release tags with Sigstore. It includes breaking changes. ## Prepare web content for Amazon Bedrock Knowledge Bases with Crawl4AI DevFeed: [Prepare web content for Amazon Bedrock Knowledge Bases with Crawl4AI]() Original publisher: [Read original article]() Author: Daxeshkumar Patel Published: 2026-11-02T17:11:21Z Content type: article Language: en Sources: [Artificial Intelligence]() Topics: [Amazon Bedrock Knowledge Bases](), [Amazon S3](), [AI Chat]() Tags: [advanced-300](), [amazon](), [amazon-bedrock](), [amazon-bedrock-knowledge-bases](), [amazon-s3](), [bedrock](), [content](), [crawler](), [custom](), [data](), [handling](), [how-to](), [javascript](), [knowledge](), [metadata](), [open](), [open-source](), [post](), [s3](), [technical-how-to](), [web](), [with]() ### AI overview For most public websites, the article recommends starting with the managed Amazon Bedrock Web Crawler. When pages need selective extraction or custom metadata, a scheduled AWS Fargate task can use Crawl4AI to crawl bounded, authorized public sites, prepare Markdown and matching metadata sidecars, upload them to S3, and start knowledge base ingestion. Bedrock then handles parsing, chunking, embeddings, indexing, and retrieval. The sample's pruning is heuristic and should be reviewed; chunking choices affect retrieval and cannot be changed on an existing data source. The workflow is batch ingestion, not real-time web access, and does not handle authenticated sessions. Its sample lacks retries and failure alarms, does not remove stale S3 objects automatically, and can incur ongoing infrastructure charges. The article advises validating retrieval and citations with representative content and treating crawled pages as untrusted model context. ### Source excerpt This post shows how to use Crawl4AI, an open source web crawler, as a content preparation layer for Amazon Bedrock Knowledge Bases, handling JavaScript rendering, selective extraction, and custom metadata enrichment before web content reaches an Amazon S3 data source. ## File Mover Express: AWS Releases Open-Source Tool for S3 File Transfers DevFeed: [File Mover Express: AWS Releases Open-Source Tool for S3 File Transfers]() Original publisher: [Read original article]() Author: Renato Losio Published: 2026-11-01T06:32:00Z Content type: release Language: en Sources: [InfoQ]() Topics: [Open Source](), [MEAN](), [Linux]() Tags: [ai](), [ai-assistants](), [amazon-ec2](), [angular](), [apache-2-0](), [application](), [aws](), [aws-identity-and-access-management-iam](), [aws-s3-file-transfer](), [cli](), [cloud](), [data-center](), [data-storage](), [data-transfer-objects](), [development](), [devops](), [go-language](), [linux](), [macos](), [news](), [open-source](), [parallel](), [performance-tuning](), [resume](), [retries](), [s3](), [verification](), [windows]() ### AI overview File Mover Express is an open-source AWS tool for moving large files between local systems and Amazon S3. Its daemon-based engine can keep transfers running independently of a desktop session, controlled through a GUI or scriptable CLI over an encrypted, password-protected connection. It supports uploads and downloads, splits large files into parallel multipart transfers, and can autotune stream counts and chunk sizes. Users can configure retries, pause and resume transfers, limit bandwidth, and enable S3 Transfer Acceleration for distant buckets. Background checksums can use MD5, XXHash, XXHash64, or XXH3, with local caching when files are unchanged. Security features include HTTPS, encryption in flight and at rest, encrypted control-plane communication, IAM bucket access, and OIDC single sign-on. Packages are available for macOS, Windows, and Linux. ### Source excerpt AWS has recently released File Mover Express, an open-source application under the Apache 2.0 license for transferring large files between local systems and Amazon S3. Available for macOS, Windows, and Linux, it supports parallel transfers, automatic performance tuning, retries, pause and resume, and file integrity verification. By Renato Losio ## Reducing petabyte-scale S3 storage costs with evidence-based lifecycle rules DevFeed: [Reducing petabyte-scale S3 storage costs with evidence-based lifecycle rules]() Original publisher: [Read original article]() Author: Marcelo Corazim Published: 2026-10-29T21:51:20Z Content type: tutorial Language: en Sources: [AWS Storage Blog]() Topics: [Amazon S3](), [cost-optimization]() Tags: [advanced-300](), [amazon-s3](), [amazon-s3-glacier-storage-classes](), [amazon-s3-intelligent-tiering](), [amazon-simple-storage-service-s3](), [analysis](), [archive](), [aws](), [aws-cloud-storage](), [best-practices](), [cost](), [costs](), [evidence-based](), [lens](), [lifecycle](), [petabyte](), [retrieval](), [rules](), [s3](), [s3-storage-lens](), [storage](), [storage-class-analysis](), [technical-how-to]() ### AI overview The article's method is to identify cold, transition-eligible S3 data with S3 Storage Lens, then observe candidate prefixes with S3 Storage Class Analysis for at least 30 days (the author used over 45 days to catch periodic reads). Size and age matter: analysis omits objects at or below 128 KB, and Glacier Instant Retrieval has a 128 KB minimum billable size and a 90-day minimum storage duration. The analysis chiefly informs Standard-to-Standard-IA transitions; for data with near-zero retrieval that still needs millisecond access, the article proposes Glacier Instant Retrieval, which has higher retrieval and request charges. Set a conservative lifecycle age based on observed access, and guard against costly scans with query routing and request-metric alarms. One engagement reports roughly 70% lower monthly storage cost for 5.8 PB while preserving millisecond access; the article cautions results depend on workload. ### Source excerpt Learn a repeatable, evidence-based method to cut petabyte-scale Amazon S3 storage costs. Use S3 Storage Lens to find untiered buckets with near-zero retrieval, prove the cold access pattern with S3 Storage Class Analysis, and move the data to S3 Glacier Instant Retrieval with a lifecycle rule while keeping millisecond access. ## Adding date-based pagination to my site DevFeed: [Adding date-based pagination to my site]() Original publisher: [Read original article]() Author: Jamie Tanna Published: 2026-10-28T08:36:51Z Content type: article Language: en Sources: [Jamie Tanna]() Topics: [Hugo](), [static-site-generator]() Tags: [adapters](), [aws](), [calendar](), [claude-opus-5-5](), [content](), [hugo](), [indieweb](), [operational](), [pagination](), [s3](), [www-jvt-me]() ### AI overview Date-based pagination in Hugo makes it possible to browse posts by month, such as visiting `/posts/2026/10/` to see October entries. For a site with roughly 46,000 mixed IndieWeb posts, the author says ordinary paginated feeds can shift on each new post and trigger rebuilds of potentially about 1 GB of HTML for deployment to AWS S3. The author says Hugo Content Adapters made date-based pagination possible and also enabled a heatmap-like calendar of posting frequency, with further tweaks still planned. ### Source excerpt 🤖 This post includes some LLM-derived content 🤖 Something I've long wanted is the ability to paginate this Hugo website's pages by day. Given this is an IndieWeb website, I post a mix of content from articles, likes of others' posts and my daily step counts, totalling roughly 46000 posts. Although this may make it easier to browse, it's also an operational want - every time I post something new, each of the feeds on my site that are paginated will move forward one entry, leading to effectively all of my feed pages re-building, leading to potentially ~1GB of HTML re-building and deploying up to AWS S3, which is not ideal! Additionally, having date-based pagination allows someone to "URL hack", by i.e. taking this blog post's URL, https://www.jvt.me/posts/2026/10/28/date-based-pagination/, and browsing to /posts/2026/10/ to see what I've written in October. This is something I've wanted as early as 2020, but has been an ask since 2014 upstream, but thankfully it's now possible using Content Adapters. This also allowed me to get Claude Opus 5.5 to create a heatmap-ish calendar for how frequently posts exist i.e. from my home page: Hopefully this is useful - I've still got some tweaks to do, and feedback welcome! ## Building document-investigation agents with Amazon Bedrock AgentCore and Amazon S3 Files DevFeed: [Building document-investigation agents with Amazon Bedrock AgentCore and Amazon S3 Files]() Original publisher: [Read original article]() Author: Ben Freiberg Published: 2026-10-27T17:16:54Z Content type: article Language: en Sources: [AWS Compute Blog]() Topics: [Amazon Bedrock AgentCore](), [Amazon S3](), [Filesystems](), [Amazon Bedrock]() Tags: [advanced-300](), [agentcore](), [agentcore-runtime](), [agentic](), [agents](), [ai](), [ai-agent](), [amazon](), [amazon-bedrock](), [amazon-bedrock-agentcore](), [amazon-bedrock-data-automation](), [amazon-bedrock-guardrails](), [amazon-cloudwatch](), [amazon-s3](), [amazon-simple-storage-service-s3](), [aws-lambda](), [bedrock](), [bedrock-agentcore](), [file-system](), [files](), [investigation](), [s3](), [technical-how-to]() ### AI overview An AWS Lambda function and an Amazon Bedrock AgentCore agent can investigate related documents through the same Amazon S3 Files mount: Lambda stages each case and a manifest, while the agent follows document references and writes findings using ordinary file operations. The design relies on mount reads for current data because exports to the S3 bucket are asynchronous; it assigns one writer per path and separate output files per producer to avoid conflicts. The article also recommends monitoring PendingExports, enforcing both IAM and POSIX permissions, and treating documents as untrusted input. Its controls include binding the case ID to the invocation, restricting paths to that case, separating intake and write locations, requiring cited evidence and review, screening tool-read text and findings with ApplyGuardrail, and logging path access. It contrasts this open-ended workflow with Amazon Bedrock Data Automation, which extracts known fields from documents and other media through a managed pipeline. ### Source excerpt Learn how an AWS Lambda function and an Amazon Bedrock AgentCore agent investigate a folder of related documents together through a shared Amazon S3 Files file system, using ordinary file operations instead of object download and upload code. ## 20 AWS Cost Optimization Moves That Still Pay Off DevFeed: [20 AWS Cost Optimization Moves That Still Pay Off]() Original publisher: [Read original article]() Author: Fedir Kompaniiets Published: 2026-10-26T15:58:42Z Content type: tutorial Language: en Sources: [HackerNoon]() Topics: [cloud cost management](), [Amazon Web Services](), [Amazon EC2](), [Graviton](), [Amazon Elastic Kubernetes Service](), [Amazon Elastic Block Store (Amazon EBS)](), [Amazon VPC](), [Amazon CloudWatch Logs](), [AWS Fargate](), [AWS consulting]() Tags: [2](), [2017](), [2025](), [amazon-s3-intelligent-tiering](), [aws](), [aws-compute-optimizer](), [aws-cost-optimization](), [aws-graviton](), [cloud-cost-optimization](), [cloud-infrastructure](), [cost](), [cost-optimization](), [ec2](), [eks](), [examples](), [finops](), [finops-cloud-optimization](), [net](), [networking](), [optimization](), [plan](), [practical](), [pricing](), [rds](), [s3](), [strategies](), [with]() ### AI overview The article argues that routine AWS defaults can drive substantial waste and recommends measuring costs first, then targeting low-effort fixes. It compares commitment options: Compute Savings Plans trade discount depth for flexibility, while EC2 Instance Savings Plans and stable database Reserved Instances can discount more; Database Savings Plans add flexibility across listed data services at a lower ceiling. Other measures include Graviton migration, Spot for interruptible work, EKS node consolidation, rightsizing, S3 tiering and lifecycle rules, gp2-to-gp3 conversion, and VPC endpoints to reduce NAT processing. Each has limits: Spot can be reclaimed, Intelligent-Tiering charges to monitor objects and is less suited to tiny objects, and interface endpoints have per-AZ and data charges. The article also recommends reducing public IPv4 use, tuning RDS and connections, setting log retention, scoping AWS Config, and assigning cost ownership through tags, budgets, alerts, and review. It emphasizes checking business impact and current regional prices; quoted savings are estimates, not additive guarantees. ### Source excerpt Discover 20 practical AWS cost optimization strategies for EC2, EKS, S3, RDS, and networking, with pricing examples and a 30-day action plan. ## StemBind: When MLLMs Get Lost Between Rules and Instances in Abstract Visual Reasoning DevFeed: [StemBind: When MLLMs Get Lost Between Rules and Instances in Abstract Visual Reasoning]() Original publisher: [Read original article]() Author: Xixiang He, Baiqi Wu, Xingming Li, Ao Cheng, Qiyao Sun, Xuanyu Ji, Qingyong Hu Published: 2026-10-26T04:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [llm-reasoning](), [Artificial Intelligence]() Tags: [benchmark](), [cs-ai](), [cs-cv](), [mapping](), [matching](), [multimodal](), [reasoning](), [s3]() ### AI overview StemBind tests abstract visual reasoning at three stages on the same image stem--perception, rule identification, and choosing the completing option--so errors can be localized instead of reduced to a single score. Across 2,298 stems and 24 MLLM configurations, rule accuracy exceeded full-answer accuracy for 22 models; even when perception and rule answers were both correct, models chose the wrong option 51.2% of the time. The authors' diagnostics locate the main difficulty in mapping an identified rule to a specific instance (S3). Larger models and explicit thinking mode did not reliably close the gap; thinking lowered rule and full-item accuracy. These findings are reported for the benchmark and configurations described in the abstract. ### Source excerpt arXiv:2606.00148v2 Announce Type: replace-cross Abstract: Multimodal large language models (MLLMs) often know the rule but pick the wrong answer: on abstract visual reasoning (AVR) tasks, a model can describe what it sees and name the underlying pattern, yet still fail to choose the matching candidate. Existing AVR benchmarks cannot detect this because they collapse perception, rule induction, and answer selection into a single right-or-wrong signal. We introduce StemBind, a shared-stem diagnostic benchmark that probes the same visual stem with three aligned questions: Perception (what is in the image), Rule (what pattern governs it), and Full (which option completes it), so a final-answer error can be attributed to a specific sub-step on the same evidence. StemBind contains 2,298 curated knowledge-light stems across nine auditable visual operations, totaling 19,533 P/R/F tasks, with each full item annotated by Sternberg's four reasoning stages (S1 Encode, S2 Infer, S3 Map, S4 Apply). Evaluating 24 frontier MLLM configurations yields four findings. (i) The R-F chasm: rule accuracy exceeds full-item accuracy on 22 of 24 models, so most failures happen after the rule is identified. (ii) A persistent binding gap: even when P and R are both correct on the same stem, models still answer F incorrectly 51.2% of the time. (iii) The bottleneck is S3: process diagnostics and Stage-wise Stimulus Augmentation localize the dominant failure to rule-to-instance mapping. (iv) Scaling and thinking do not help: neither larger models nor explicit thinking mode reliably closes the gap, and thinking even lowers rule and full-item accuracy. StemBind reframes AVR evaluation from final-answer ranking to locating where abstract visual reasoning breaks down, identifying rule-to-instance binding as a concrete next target for vision-grounded reasoning. ## CSIRO Uses Serverless Architecture to Query 100x More Genomic Variants on AWS DevFeed: [CSIRO Uses Serverless Architecture to Query 100x More Genomic Variants on AWS]() Original publisher: [Read original article]() Author: Leela Kumili Published: 2026-10-23T14:30:00Z Content type: news Language: en Sources: [InfoQ]() Topics: [Serverless](), [Cloud Architecture](), [AWS Lambda](), [Amazon Web Services](), [Amazon DynamoDB]() Tags: [amazon](), [amazon-s3](), [applied-research](), [architecture](), [architecture-design](), [athena](), [aws](), [aws-lambda](), [base](), [beacon](), [central](), [cloud-architecture](), [cost-optimization](), [csiro-serverless-genomic-variant](), [database](), [development](), [distributed-query](), [distributed-systems](), [dynamodb](), [files](), [http](), [indexes](), [lambda](), [latency](), [metadata](), [news](), [query](), [s3](), [scalability](), [science](), [serverless](), [serverless-architecture](), [uses]() ### AI overview Serverless Beacon lets CSIRO query genomic variants across cohorts while VCF files remain in Amazon S3 instead of being copied into a central database. It keeps searchable metadata in Athena and DynamoDB indexes, with Ontoserver supporting ontology terms for queries that combine genomic and phenotypic information. API Gateway routes requests to Lambda; the query module uses bcftools/htslib HTTP byte-range requests to read only relevant regions of compressed VCF files, then aggregates results through the Beacon API. In an evaluation using chromosome 1 from the 1000 Genomes Project (2,504 samples; 1.1 GB compressed VCF), onboarding took 18 seconds. Queries spanning 10,000 base pairs averaged 1.52 seconds and an estimated USD 0.00013 in compute cost; latency stayed broadly consistent as matching variants rose from four to 400. These are reported results for that evaluation dataset. ### Source excerpt CSIRO has developed Serverless Beacon, a serverless architecture for querying genomic variants without copying VCF files into a central database. The system combines Amazon S3, Lambda, Athena and DynamoDB, using metadata indexes and HTTP byte range requests to query compressed genomic regions. Tests reported 1.52 second average latency for 10,000 base pair queries. By Leela Kumili ## How Seltz cut data costs 40% with Amazon S3 Tables DevFeed: [How Seltz cut data costs 40% with Amazon S3 Tables]() Original publisher: [Read original article]() Author: Vivien de Saint Pern Published: 2026-10-22T19:47:31Z Content type: article Language: en Sources: [AWS Storage Blog]() Topics: [Amazon S3 Tables](), [Apache Iceberg](), [Amazon S3](), [data-engineering]() Tags: [amazon](), [amazon-s3](), [amazon-s3-tables](), [apache](), [apache-iceberg](), [automatic](), [compaction](), [cost](), [costs](), [customer-solutions](), [cut-data](), [data](), [data-costs](), [iceberg](), [incremental](), [infrastructure](), [knowledge](), [s3](), [s3-tables](), [tables](), [with]() ### AI overview Seltz reports 40% lower infrastructure cost with Amazon S3 Tables than with its self-managed pipeline, based on cost modeling at its current scale; the figure excludes engineering effort. Its Knowledge Pipeline writes each enrichment stage's intermediate results to a separate Iceberg table, and a downstream indexer builds search index files. S3 Tables automatically compacts small writes and removes expired or unreferenced data, while atomic commits, independent table access, and file-level pruning support incremental processing. Seltz uses Athena for SQL monitoring and DuckDB for local inspection. The article also outlines a walkthrough using PyIceberg and Athena, requiring an AWS account with table-bucket permissions, Python 3.9+, and configured Athena. ### Source excerpt Learn how Seltz uses Amazon S3 Tables as managed Apache Iceberg storage between the stages of its billion-document Knowledge Pipeline. The team gets automatic compaction, lifecycle management, and incremental table semantics without operating that infrastructure, cutting infrastructure cost 40 percent compared to a self-managed pipeline. ## WaRP: Workload-Aware Request Planning for Vector Retrieval on Object Storage DevFeed: [WaRP: Workload-Aware Request Planning for Vector Retrieval on Object Storage]() Original publisher: [Read original article]() Author: Ruipeng Wan, Xinyu Zeng, Yiming Qiao, Huanchen Zhang Published: 2026-10-22T04:00:00Z Content type: article Language: en Sources: [arXiv: Databases]() Topics: [ann](), [Amazon S3](), [Vectors & LLMs]() Tags: [ann](), [aws](), [concurrency](), [cs-db](), [latency](), [object-storage](), [optimization](), [planning](), [request](), [s3](), [scheduler](), [vector]() ### AI overview WaRP plans how to merge vector-cluster byte ranges into object-storage requests, reducing per-request overhead and billed GETs while preserving ANN recall. It uses sorted intervals, a per-query request limit, and an estimate of request concurrency to optimize a latency surrogate; its dynamic program and adaptive frontier search return a plan within (1+ε) of optimal, with planning under 10 ms in the most demanding tested settings. An offline reordering pass improves mergeability without changing ANN semantics. In an AWS S3 evaluation integrated with Lance's I/O scheduler across seven IVFPQ datasets, WaRP plus reordering achieved 1.09-3.48x higher QPS and 2.12-5.19x fewer GETs than Lance, with the same recall. ### Source excerpt arXiv:2610.29689v1 Announce Type: new Abstract: Modern vector search systems keep payload vectors in cloud object storage and only a compact routing index on fast local storage, typically using cluster-based ANN indexes. This design inherits a problem existing systems do not address. On object storage, small Range-GETs have latency dominated by fixed per-request overhead and are billed per request. Existing readers either issue one request per probed cluster, or coalesce adjacent ranges with hand-tuned gap or span thresholds that do not transfer across datasets or query workloads. We formulate the grouping of probed-cluster byte intervals into object-store requests as a global per-query optimization problem and present WaRP, a workload-aware request planner that solves it efficiently. Given the sorted byte intervals, a limit on the number of requests per query, and an estimate of how many object-store requests from one query can proceed concurrently under the target serving configuration, WaRP searches for a merge plan that minimizes a concurrency-aware latency surrogate while satisfying the request limit. Using an optimized dynamic program and adaptive frontier search, WaRP returns a plan within $(1{+}\epsilon)$ of the optimum, with per-query planning overhead under $10$ms even in the most demanding tested settings. A lightweight offline reordering pass further improves request mergeability without changing the semantics of ANN search. We implement WaRP inside Lance's I/O scheduler and evaluate it on AWS S3. We build IVFPQ indexes on seven datasets, including SIFT100M and DEEP100M at the 100M-vector scale. WaRP + Reorder improves QPS by $1.09--3.48\times$ and reduces GETs by $2.12--5.19\times$ relative to Lance, while preserving Lance's recall. ## How Proscia built unlimited observability with Amazon S3 and the Altinity Kubernetes Operator DevFeed: [How Proscia built unlimited observability with Amazon S3 and the Altinity Kubernetes Operator]() Original publisher: [Read original article]() Author: Kyriakos Toulgaridis Published: 2026-10-21T16:40:04Z Content type: article Language: en Sources: [AWS Storage Blog]() Topics: [observability](), [Amazon S3](), [Amazon Elastic Kubernetes Service](), [observability pipelines](), [Self-hosted](), [Amazon CloudWatch](), [LLM observability](), [agent observability]() Tags: [advanced-300](), [ai](), [amazon](), [amazon-athena](), [amazon-cloudwatch](), [amazon-cognito](), [amazon-eks](), [amazon-elastic-block-storage-amazon-ebs](), [amazon-elastic-block-store-amazon-ebs](), [amazon-elastic-kubernetes-service](), [amazon-elastic-kubernetes-service-amazon-eks](), [amazon-s3](), [amazon-simple-storage-service-amazon-s3](), [amazon-simple-storage-service-s3](), [aws-cloud-storage](), [aws-identity-and-access-management-iam](), [aws-key-management-service](), [aws-key-management-service-aws-kms](), [can](), [clickhouse](), [company](), [cost](), [customer-solutions](), [data](), [data-residency](), [eks](), [gdpr](), [hipaa](), [it](), [kubernetes](), [kubernetes-operator](), [observability](), [operator](), [s3](), [with]() ### AI overview Proscia built a self-managed observability platform across AWS accounts and Regions: OpenTelemetry collectors send telemetry to a dedicated account, where ClickHouse runs on EKS under the Altinity Kubernetes Operator. Recent data stays on EBS SSDs; TTL rules move colder data to S3, and backups also go to S3. The design combines high-cardinality queries, long retention, and healthcare compliance controls, with Grafana for dashboards and alerting and an MCP interface for natural-language queries. Proscia reports handling 33 billion records monthly, over 10 times more data at roughly the same overall cost as its prior SaaS service, 100 times lower storage cost per TB, and median root cause analysis in 10 minutes versus a prior 4-24-hour escalation round trip. Building requires engineering capacity to operate stateful Kubernetes systems; the text suggests a hybrid approach using managed ClickHouse or hosted visualization when that is a poor fit. ### Source excerpt Observability can claim a growing share of IT budgets. Learn how Proscia, a pathology AI company, built a self-hosted observability platform on Amazon S3, Amazon EKS, and ClickHouse managed by the Altinity Kubernetes Operator, achieving unlimited telemetry retention at object-storage cost under HIPAA, GDPR, and data-residency requirements. ## Updating encryption on existing Amazon S3 objects with S3 Batch Operations DevFeed: [Updating encryption on existing Amazon S3 objects with S3 Batch Operations]() Original publisher: [Read original article]() Author: Rohit Rajput Published: 2026-10-21T16:01:22Z Content type: tutorial Language: en Sources: [AWS Storage Blog]() Topics: [Amazon S3](), [secret rotation]() Tags: [advanced-300](), [amazon-s3](), [amazon-s3-batch-operations](), [amazon-s3-intelligent-tiering](), [amazon-s3-lifecycle](), [amazon-s3-object-lock](), [amazon-simple-storage-service-amazon-s3](), [amazon-simple-storage-service-s3](), [api](), [aws-cloud-storage](), [aws-identity-and-access-management-iam](), [aws-key-management-service](), [aws-key-management-service-aws-kms](), [batch](), [data-movement](), [encryption](), [kms](), [lifecycle](), [s3](), [scale](), [technical-how-to](), [updating]() ### AI overview The excerpt describes using S3 Batch Operations with the `UpdateObjectEncryption` API to change existing SSE-S3 objects to SSE-KMS in place. After changing the bucket's default encryption for future writes, create a job filtered to SSE-S3 objects, choose a customer managed KMS key and optionally enable S3 Bucket Keys, then review and run the job. A completion report records successes, failures, and skipped objects. The update avoids copies and data movement, preserves object properties and Lifecycle eligibility, and applies across object sizes and storage classes in general purpose buckets. Limits and operational requirements include S3 Object Lock preventing updates in some cases, no support for unencrypted, DSSE-KMS, or SSE-C objects, and the need to update source and destination separately for replication. Readers need `kms:Decrypt` access to read the newly encrypted objects; Bucket Keys can reduce KMS charges by up to 99 percent. The excerpt also warns that replication roles need appropriate KMS permissions before changing default encryption. ### Source excerpt Large organizations accumulate billions of Amazon S3 objects under different encryption configurations. Learn how to use S3 Batch Operations with the new UpdateObjectEncryption API to standardize encryption across an entire bucket at scale. The approach moves objects from SSE-S3 to SSE-KMS in place, with no data movement and without breaking S3 Lifecycle policies. ## What requirements don't you have? DevFeed: [What requirements don't you have?]() Original publisher: [Read original article]() Published: 2026-10-19T00:00:00Z Content type: opinion Language: en Sources: [Sean Goedecke]() Topics: [Requirements](), [Cache](), [Go Language](), [Git]() Tags: [cache](), [concurrency](), [design](), [distributed](), [isolation](), [overhead](), [requirements](), [s3](), [service](), [simplicity](), [sockets](), [speed](), [sqlite](), [stats](), [storage](), [system](), [system-design](), [udp](), [unicorn](), [update]() ### AI overview Elegant software design often comes from explicitly relaxing requirements: Unicorn gains process isolation and simpler request handling while relying on a proxy for slow clients; S3 trades atomic multi-key updates for faster, more scalable blob-level storage; StatsD accepts UDP data loss to reduce network overhead; edge caches permit staleness for speed; SQLite gives up concurrent writes for operational simplicity; and Cursor accepts high tail latency to scale its distributed Git backend. The text argues that deciding which compromises users can accept requires both technical and business judgment. It also warns that coding agents may invent edge-case requirements and add unnecessary locks, checks, and coordination, making changes and systems more complex. The author recommends stating acceptable staleness or inconsistency explicitly and reviewing generated code for unjustified synchronization. ### Source excerpt When designing software, the requirements you don't have are as important as the requirements you do. Identifying what you can compromise on -- what can be slow, or stale, or inflexible -- is usually the key to an elegant design. Elegant systems drop some requirements For instance, take preforking web servers like Unicorn. Because they lack the feature of shared mutable state between server instances, they can run each server in its own process, which provides isolation for free. Because Unicorn specifically ignores the problem of slow client handling1, it can queue requests directly on Unix sockets, drastically simplifying the implementation. Because S3 does not support atomic multi-key updates -- you cannot update multiple blobs in a single transaction -- it can store data at the blob level however it chooses. That makes it much faster, since reads and writes don't have to cross storage partitions, and much easier to scale, since less coordination means fewer sources of truth. StatsD uses UDP by default, which means it doesn't guarantee that every piece of data you send makes it into storage. In exchange for dropping that requirement, it incurs much less network overhead, which is a big advantage for a stats aggregation service that's constantly shoveling data over the wire. The general design approach of fanning out data to an edge cache buys a lot of speed and scaling by dropping the requirement that data must always be up-to-date. SQLite buys operational simplicity -- it can run out of a single file! -- by dropping the requirement of concurrent writes. By accepting high tail latency2, Cursor can scale their distributed git backend more easily. Well-designed systems turn their weaknesses into strengths. Why dropping requirements is hard Anyone who's spent time designing real-world software (or who's studied distributed systems) knows that system design is a process of compromise. If you attempt to satisfy every single potential requirement -- speed, concurrency, up-to- ## Writing to AWS S3 from Fly.io (in Go) DevFeed: [Writing to AWS S3 from Fly.io (in Go)]() Original publisher: [Read original article]() Author: Jamie Tanna Published: 2026-10-18T16:46:39Z Content type: tutorial Language: en Sources: [Jamie Tanna]() Topics: [AWS Security Token Service](), [fly.io](), [OpenID connect (OIDC)](), [IO]() Tags: [aws](), [blogumentation](), [fly](), [fly-io](), [go](), [oidc](), [s3]() ### AI overview A Go tutorial shows how to obtain a short-lived OpenID Connect token from Fly.io, exchange it through AWS STS for temporary credentials, and use those credentials to write objects to S3. It also gives example IAM trust and bucket policies and notes that overly restrictive trust-policy conditions caused an access-denied error. ### Source excerpt I'm currently in the process of moving my site's media (under media.jvt.me) from Netlify to AWS S3. The underlying infrastructure for my site that writes this media is hosted on Fly.io, so needs a way to write to S3. I really wanted to avoid long-lived AWS credentials, and fortunately Fly supports OpenID Connect which integrates with AWS' OpenID Connect Federation to allow you to use short-lived OpenID Connect tokens to authenticate as a given IAM role, and then perform any necessary AWS actions. Although Fly has an example with curl, I wanted to blogument the "how to" with Go, as it wasn't super straightforward to do so. I've written this up in a sample repo on GitLab.com. Go code The core pieces are: // via https://gist.github.com/teknoraver/5ffacb8757330715bcbcc90e6d46ac74 var unixClient = http.Client{ Transport: &http.Transport{ DialContext: func(_ context.Context, _, _ string) (net.Conn, error) { return net.Dial("unix", "/.fly/api") }, }, } func getOIDCToken(unixClient http.Client) (string, error) { s3Body := bytes.NewReader([]byte(`{"aud":"sts.amazonaws.com"}`)) // or could be `{"aud":"https://sts.amazonaws.com"}` as long as it matches your Trust Policy // get the OIDC token resp, err := unixClient.Post("http://localhost/v1/tokens/oidc", "application/json", s3Body) if err != nil { slog.Error("Failed to call Fly OIDC tokens endpoint", "err", err) return "", err } defer resp.Body.Close() body, err := io.ReadAll(resp.Body) if err != nil { slog.Error("Failed to read body", "err", err) return "", err } return string(body), nil } func newAWSClient(oidcToken string) (*s3.Client, error) { cfg, err := config.LoadDefaultConfig(context.TODO(), config.WithRegion("eu-west-1")) if err != nil { return nil, err } stsClient := sts.NewFromConfig(cfg) assumedRoleInput := &sts.AssumeRoleWithWebIdentityInput{ RoleArn: aws.String(os.Getenv("AWS_ASSUME_ROLE_ARN")), RoleSessionName: aws.String("..."), WebIdentityToken: aws.String(oidcToken), } resp, err := stsClient.AssumeRoleWithWeb ## Amazon Acquires DuckLabs and Integrates DuckDB in Amazon Aurora PostgreSQL DevFeed: [Amazon Acquires DuckLabs and Integrates DuckDB in Amazon Aurora PostgreSQL]() Original publisher: [Read original article]() Author: Gianmarco Nalin Published: 2026-10-17T06:00:00Z Content type: news Language: en Sources: [InfoQ]() Topics: [DuckDB](), [PostgreSQL](), [Open Source](), [Amazon S3](), [Amazon Web Services](), [Amazon S3 Tables]() Tags: [acquisition](), [acquisitions](), [ai](), [ai-ml-data-engineering](), [amazon](), [amazon-acquires-ducklabs](), [amazon-aurora](), [amazon-rds](), [amazon-s3](), [amazon-web-services](), [apache-iceberg](), [apache-parquet](), [architecture-design](), [aws](), [cloud](), [data-analytics](), [data-lake](), [development](), [duckdb](), [ducklake](), [ml-data-engineering](), [news](), [open-source](), [postgresql](), [s3]() ### AI overview Amazon acquired DuckLabs, the company behind DuckDB, with its co-founders continuing to lead the team and open-source project's technical direction at AWS. DuckDB and the Duck Stack will remain free and open source under the MIT license, with the DuckDB Foundation continuing project oversight. The article also describes a DuckDB-powered feature in Aurora PostgreSQL that lets queries combine operational data with historical data in Amazon S3 and related storage. ### Source excerpt Amazon has acquired DuckLabs, the company behind the analytical database DuckDB. Co-founders Hannes Mühleisen and Mark Raasveldt will continue to lead the team under AWS. The acquisition aims to enhance the development and reach of DuckDB while maintaining its open-source status. Amazon Aurora PostgreSQL gets its first DuckDB-powered feature following the acquisition. By Gianmarco Nalin ## Stop renting access to your own files DevFeed: [Stop renting access to your own files]() Original publisher: [Read original article]() Author: Ahson Shaikh Published: 2026-10-16T09:12:31Z Content type: article Language: en Sources: [DevOps.dev]() Topics: [Self Hosted File Indexing and Searching](), [dropbox](), [on-prem]() Tags: [business](), [drive](), [dropbox](), [files](), [google](), [google-drive](), [google-one](), [object-storage](), [onedrive](), [r2](), [s3](), [self-hosted](), [self-hosting](), [server](), [storage](), [yourdrive]() ### AI overview yourdrive is a self-hosted web drive that puts a familiar file interface over object storage in an account the user controls. It runs in Docker on a Linux server and supports AWS S3, Azure Blob Storage, Cloudflare R2, Wasabi, MinIO, and other S3-compatible services. Browsers upload directly to the bucket, so files do not use the server's disk. The article contrasts its per-server licensing with per-user plans: Community is free for up to 3 users, Team costs $119/year for up to 25, and Business is listed at $599/year for up to 250, with SAML SSO, audit log, and groups described as coming soon. It presents this as a lower-cost option for small teams with substantial storage, while acknowledging that a household under Google One's 2 TB cap may pay less with Google. Users must maintain a Linux server; the product is a web app rather than a desktop sync client or office suite. ### Source excerpt Why I built yourdrive, a self-hosted drive that keeps your photos, videos and documents in a bucket you already own. Count the drives you pay for. A personal Google plan for the family photos. A Dropbox seat for every person on your team. Maybe a OneDrive that came bundled with something else and now holds files you forgot about. Each one charges you every month, per person, for access to files you made. Stop paying and the files sit behind a login you no longer control. That is not storage. That is rent. I built yourdrive because I wanted the opposite: a drive that looks and feels like the one you already use, but runs on your own server and keeps every file in your own bucket. No per-seat fees. No middleman holding the keys. A drive that's actually yours. The quiet problem with rented drives Google Drive is fine for one household. Extra storage comes through Google One, at $99.99 a year for 2 TB. But that is a personal plan with a hard cap, and the moment a team needs a shared drive, the pricing switches to seats. Team drives are priced per person, not per file. Dropbox Standard costs $15 per user per month. Box Business costs $15 per user per month on an annual plan. Hire someone, and the bill goes up whether they upload one file or ten thousand. That model hides three costs: The bill tracks headcount, not storage. A 25-person team pays $4,500 a year on either plan, before anyone uploads a file. Your files live under someone else's terms. Prices, limits and features change on their schedule. You adapt. Leaving is a project. Moving years of photos and documents out is slow enough that most people never do it. Meanwhile, plenty of teams already pay for object storage: an S3 bucket for backups, an R2 bucket behind a website. It is cheap, durable and in their own account. What it lacks is a friendly drive on top, one a designer or an accountant can use without touching a terminal. What yourdrive is yourdrive is a self-hosted drive for photos, videos and documents. Yo ## Announcing the official Appwrite Helm chart for Kubernetes DevFeed: [Announcing the official Appwrite Helm chart for Kubernetes]() Original publisher: [Read original article]() Author: Levi van Noort Published: 2026-10-15T00:00:00Z Content type: article Language: en Sources: [Appwrite Blog]() Topics: [Helm charts](), [Appwrite](), [self-hosted databases vs managed services](), [Amazon Elastic Kubernetes Service]() Tags: [announcements](), [api](), [argo-cd](), [clickhouse](), [cloudnativepg](), [helm](), [kubernetes](), [postgresql](), [redis](), [s3]() ### AI overview Appwrite announces an official Helm chart for deploying and upgrading self-hosted Appwrite on Kubernetes. The chart supports built-in or managed datastores, Functions and Sites, configurable worker layouts, and migration handling. Its defaults and operational guidance address issues such as singleton jobs, resource sizing, credential persistence, and the privileged executor required for Functions and Sites. ### Source excerpt Self-host Appwrite on Kubernetes with the official Helm chart, using built-in or managed PostgreSQL, Redis, and ClickHouse, S3, and Functions and Sites. [Next page]()