# B2Cloud Published articles for B2Cloud. 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 Workloads Are Breaking Old Storage Assumptions DevFeed: [AI Workloads Are Breaking Old Storage Assumptions]() Original publisher: [Read original article]() Author: Maddie Presland Published: 2026-11-05T16:40:36Z Content type: article Language: en Sources: [Backblaze]() Topics: [AI Infrastructure](), [systems](), [data-architecture](), [GPU]() Tags: [2013](), [2022](), [2025](), [2026](), [aging](), [ai](), [ai-data](), [ai-data-centers](), [ai-ml](), [ai-workloads](), [b2cloud](), [backblaze](), [cloud](), [cloud-storage](), [data](), [data-movement](), [drive](), [featured](), [featured-cloud-storage](), [network](), [performance](), [reliability](), [stats](), [storage]() ### AI overview Storage evaluation for AI workloads needs more than price per terabyte: bursty, parallel data movement and weak performance can leave costly GPUs idle. Backblaze presents three data sources: Network Stats tracks traffic by network type and reports unusually high neocloud traffic, while noting the dataset is young and limited to Backblaze's own network; Performance Stats benchmarks single- and multithreaded uploads and downloads across file sizes, publishing methodology and rate-limited results, but says two quarters are too little for firm conclusions; Drive Stats reports long-term hard-drive reliability, not SSDs, with a 2025 annual failure rate of 1.36% across 557 million drive-days. The post recommends using these findings to map requirements to workload needs and running tests against one's own stack. It also describes how larger drives increase storage density while adding complexity to erasure-coded protection. ### Source excerpt AI workloads are rewriting the rules for cloud storage. Explore what Backblaze's Network Stats, Performance Stats, and Drive Stats reveal about data movement, storage performance, and drive reliability, and use the findings to evaluate storage for your own AI workloads. The post AI Workloads Are Breaking Old Storage Assumptions appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Scaling AI Data Storage and Delivery with NVIDIA AIStore and B2 DevFeed: [Scaling AI Data Storage and Delivery with NVIDIA AIStore and B2]() Original publisher: [Read original article]() Author: Jeronimo De Leon Published: 2026-11-03T17:29:46Z Content type: article Language: en Sources: [Backblaze]() Topics: [AI Infrastructure](), [Cache](), [Amazon S3](), [big-data](), [data lake](), [Persistence](), [Isaac](), [sharding]() Tags: [ai](), [ai-data](), [ai-ml](), [ai-workloads](), [b2cloud](), [backblaze](), [backblaze-b2](), [capacity](), [cloud](), [cloud-storage](), [compute](), [curation](), [data](), [data-storage](), [datasets](), [delivery](), [fast](), [featured](), [featured-cloud-storage](), [local](), [nvidia](), [scaling](), [scaling-ai](), [storage]() ### AI overview Backblaze B2 holds the durable dataset record while NVIDIA AIStore caches and stages active data on storage near compute, balancing retention cost with faster access. For S3-based processing such as NeMo Curator, applications can use AIStore through configuration, prefetch data, and reuse cached reads. For Isaac GR00T training, AIStore stages a dataset revision to GPU-node disk and checks it against a manifest, leaving the trainer's data loader unchanged; checkpoints publish back through AIStore to B2. The post recommends sizing AIStore for concurrent working sets and reuse, using prefetch when predictable startup matters, keeping dataset revisions in write-once prefixes, publishing checksummed manifests, and measuring loading, warm reads, processing, and publication separately. The approach suits repeated or shared reads; a single sequential pass may be simpler streamed directly from B2. AIStore is infrastructure teams operate, and write-through adds local-write overhead before upload. ### Source excerpt AI workloads need durable data retention and fast access to active datasets. See how Backblaze B2 and NVIDIA AIStore pair a persistent capacity tier with local storage near compute for video curation, robot training, and repeated reads across shared datasets. The post Scaling AI Data Storage and Delivery with NVIDIA AIStore and B2 appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Network Stats Q3 2026: Higher Baselines, Shifting Patterns DevFeed: [Network Stats Q3 2026: Higher Baselines, Shifting Patterns]() Original publisher: [Read original article]() Author: Brent Nowak Published: 2026-10-22T13:00:00Z Content type: news Language: en Sources: [Backblaze]() Topics: [Internet Trends](), [Data analysis](), [Data visualization](), [Cloud]() Tags: [2023](), [2025](), [2026](), [about](), [absolute](), [b2cloud](), [backblaze](), [capacity](), [cloud](), [cloud-storage](), [data](), [demand](), [explore](), [featured](), [featured-cloud-storage](), [latest](), [neocloud](), [network](), [networkstats](), [patterns](), [post](), [q3](), [stats](), [storage](), [traffic](), [uncategorized]() ### AI overview Backblaze's network traffic hit a new peak in July 2026, while CDN traffic stayed above its prior baseline and neocloud demand remained high through Q3. The report finds regional shifts, including strong US-East CDN growth and rising neocloud activity across US-East, US-West, and EU-Central; it cautions that these patterns reflect Backblaze's customers and workflows and may be affected by a few high-volume flows. To handle variable demand, the network team expanded private interconnections and transit, reporting year-over-year growth of 79% on PNI links and 61% on transit, and moved provisioning from dual 100 Gbps links toward multi-100 Gbps and 400 Gbps deployments. The dataset covers three quarters of operationalized comparisons, so the authors say they will continue tracking whether trends persist. ### Source excerpt Backblaze's Q3 2026 Network Stats reveal higher traffic baselines, expanding neocloud activity, and regional shifts in data movement. Explore the latest heatmaps and see how growing demand is shaping the connections and capacity behind our increasingly busy cloud storage network. The post Network Stats Q3 2026: Higher Baselines, Shifting Patterns appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Neoclouds: Your Storage Layer Just Got a Public Grade DevFeed: [Neoclouds: Your Storage Layer Just Got a Public Grade]() Original publisher: [Read original article]() Author: David Johnson Published: 2026-10-20T14:21:23Z Content type: article Language: en Sources: [Backblaze]() Topics: [Infrastructure as a Service (IaaS)](), [deltalake](), [Amazon S3](), [BlueField DPU](), [Grafana]() Tags: [ai](), [ai-labs](), [ai-ml](), [aws](), [azure](), [b2cloud](), [backend](), [bandwidth](), [ceph](), [cloud-storage](), [cluster](), [compute](), [configuration](), [control-plane](), [coreweave](), [data-processing](), [featured](), [featured-cloud-storage](), [isolation](), [neoclouds](), [performance](), [providers](), [storage](), [strategy]() ### AI overview SemiAnalysis' ClusterMAX 3.0 requires neoclouds to provide both high-performance file and S3-compatible object storage, then tests speed and performance under 15 minutes of mixed traffic; some providers lost nearly 40% of bandwidth. The excerpt says only 19 of 77 providers tested in depth earned a Medallion rating, and describes recurring problems such as ineffective volume RBAC, exposed backend networks, and default DPU settings that can let a GPU-cloud customer reach tenant-isolation controls. It distinguishes deliberate storage design from defaults that went unchecked, citing examples ranging from missing shared filesystems to a fast Lustre tier undermined by a broken default StorageClass. Providers can build on Ceph, build their own storage hardware, or partner: these trade control and staffing or capital commitments differently. The article urges providers to decide storage responsibilities and isolation before customers leave data on the platform. ### Source excerpt SemiAnalysis' ClusterMAX 3.0 puts neocloud storage under scrutiny. From performance to tenant isolation, default configurations can undermine customer confidence. Explore what the findings reveal and why providers need a deliberate storage strategy before customers leave their data on the platform. The post Neoclouds: Your Storage Layer Just Got a Public Grade appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Backblaze Drive Stats for Q2 2026 DevFeed: [Backblaze Drive Stats for Q2 2026]() Original publisher: [Read original article]() Author: Stephanie Doyle Published: 2026-09-29T13:00:00Z Content type: article Language: en Sources: [Backblaze]() Topics: [RAID](), [Amazon S3]() Tags: [2026](), [b2cloud](), [cloud-storage](), [data](), [data-center](), [featured](), [featured-cloud-storage](), [hard-drive-stats](), [storage](), [uncategorized]() ### AI overview Backblaze analyzed 354,415 data drives for Q2 2026 and reported a 1.73% quarterly annualized failure rate and a 1.41% lifetime rate. The report reviews drive models and discusses higher-capacity drives and storage technologies relevant to future data center deployments. ### Source excerpt We analyzed 354,415 data drives for Q2 2026, revealing a 1.73% quarterly annualized failure rate and a 1.41% lifetime rate. Explore standout models, high-capacity drive deployment, and what CMR, SMR, and HAMR could mean for future data center storage strategies. The post Backblaze Drive Stats for Q2 2026 appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Why the Best AI Clouds Don't Run on Flash Alone DevFeed: [Why the Best AI Clouds Don't Run on Flash Alone]() Original publisher: [Read original article]() Author: David Johnson Published: 2026-09-24T14:31:07Z Content type: article Language: en Sources: [Backblaze]() Topics: [Artificial Intelligence](), [cloud-infrastructure](), [Cloud](), [data gravity](), [data]() Tags: [ai](), [ai-ml](), [b2cloud](), [capacity](), [cloud](), [cloud-storage](), [egress](), [featured](), [featured-cloud-storage](), [gpu](), [gpu-computing](), [neocloud](), [storage]() ### AI overview AI training requires flash storage for high-speed data delivery, but other stages of the AI lifecycle can use cheaper, higher-capacity storage. The article argues that tiered storage can reduce costs, limit hyperscaler lock-in, and help neoclouds retain broader customer relationships. ### Source excerpt AI training needs flash, but the rest of the AI lifecycle does not. See how tiered storage helps neoclouds lower costs, keep data close to GPUs, avoid hyperscaler lock-in, and retain more of the customer relationship across every workload stage. The post Why the Best AI Clouds Don't Run on Flash Alone appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Backblaze B2 x Suite Studios: S3 Native File Streaming Turns B2 Cloud Storage Into a High-Performance Drive DevFeed: [Backblaze B2 x Suite Studios: S3 Native File Streaming Turns B2 Cloud Storage Into a High-Performance Drive]() Original publisher: [Read original article]() Author: Dave Simon Published: 2026-09-10T14:04:38Z Content type: article Language: en Sources: [Backblaze]() Topics: [Amazon S3](), [Cloud](), [Streaming](), [Filesystems](), [mount](), [data](), [migration]() Tags: [architecture](), [automation](), [b2cloud](), [cloud](), [cloud-storage](), [data](), [featured](), [featured-cloud-storage](), [media](), [media-workflow](), [migration](), [mount](), [nas](), [object-storage](), [s3](), [storage](), [streaming]() ### AI overview Backblaze B2 integrates with Suite Studios' S3 Native File Streaming to provide drive-like, high-performance access to cloud-stored data. Teams can mount B2 buckets, stream only the file portions applications need, and use standard S3-compatible objects without duplicating or relocating datasets. The integration supports media production and other workflows involving large cloud datasets. ### Source excerpt Backblaze B2 now integrates with Suite Studios S3 Native File Streaming, giving teams high-performance, drive-like access to cloud data. Work directly with standard S3-compatible objects across media, scientific, geospatial, and engineering workflows without duplicating datasets or creating new storage silos. The post Backblaze B2 x Suite Studios: S3 Native File Streaming Turns B2 Cloud Storage Into a High-Performance Drive appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Introducing the Backblaze B2 MCP Server DevFeed: [Introducing the Backblaze B2 MCP Server]() Original publisher: [Read original article]() Author: Jeronimo De Leon Published: 2026-09-08T18:13:00Z Content type: article Language: en Sources: [Backblaze]() Topics: [Model Context Protocol](), [Cloud](), [Authorization](), [Open Source]() Tags: [agents](), [ai](), [ai-ml](), [api](), [b2cloud](), [cloud](), [cloud-storage](), [developers](), [featured](), [featured-cloud-storage](), [mcp](), [mcp-server](), [object-storage](), [open-source](), [storage](), [uncategorized]() ### AI overview Backblaze introduces an open-source B2 MCP Server that lets AI agents operate cloud object storage through MCP-compatible clients. Access is constrained by the permissions of the connected B2 application key, including limits on buckets and operations. ### Source excerpt The Backblaze B2 MCP Server gives AI agents a safe, standard way to operate cloud object storage. Connect MCP-compatible clients to scoped B2 application keys, manage files and buckets, move large objects directly, and control destructive actions with configurable safeguards. The post Introducing the Backblaze B2 MCP Server appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Four Platforms, One Timeline: Bringing TAMS Workflow to Life on Backblaze B2 DevFeed: [Four Platforms, One Timeline: Bringing TAMS Workflow to Life on Backblaze B2]() Original publisher: [Read original article]() Author: Dave Simon Published: 2026-09-03T14:23:57Z Content type: article Language: en Sources: [Backblaze]() Topics: [cloud-infrastructure](), [API](), [Kubernetes](), [browser](), [systems]() Tags: [api](), [article](), [b2cloud](), [browser](), [cloud](), [cloud-native](), [cloud-storage](), [featured](), [featured-cloud-storage](), [kubernetes](), [media](), [object-storage](), [storage]() ### AI overview The article explains how Time Addressable Media Store (TAMS) organizes live media by timeline-based, timestamped segments in object storage. It describes a cloud-native, interoperable workflow built around Backblaze B2 and ecosystem partners LiveWyer, CuttingRoom, and Drastic Technologies, enabling ingest, editing, distribution, playback, and archiving through shared access to the same content. ### Source excerpt TAMS organizes live media by time, giving distributed production teams faster access to shared footage. See how Backblaze B2, LiveWyer, CuttingRoom, and Drastic Technologies create an interoperable, cloud-native workflow for ingest, editing, playback, distribution, and archive with economics at scale. The post Four Platforms, One Timeline: Bringing TAMS Workflow to Life on Backblaze B2 appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Backblaze B2 To Encrypt New Uploads by Default DevFeed: [Backblaze B2 To Encrypt New Uploads by Default]() Original publisher: [Read original article]() Author: Backblaze Published: 2026-08-27T13:00:00Z Content type: release Language: en Sources: [Backblaze]() Topics: [Cloud](), [Encryption](), [Security](), [configuration]() Tags: [b2cloud](), [cloud](), [cloud-storage](), [configuration](), [encryption](), [featured](), [featured-cloud-storage](), [secure-by-default](), [security](), [storage](), [uncategorized](), [update]() ### AI overview Backblaze B2 will automatically apply server-side AES-256 encryption to newly uploaded and copied object data at rest. SSE-B2 requires no application changes, added cost, or performance tradeoff, while SSE-C remains available for objects encrypted with customer-provided keys. ### Source excerpt Starting September 14, Backblaze B2 will automatically encrypt every new upload and destination copy at rest with AES-256. Always-on SSE-B2 strengthens security without application changes, added costs, or performance impact, while preserving SSE-C for customer-managed encryption keys when required instead. The post Backblaze B2 To Encrypt New Uploads by Default appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## How Adobe Reduced GPU Idle Time in Generative AI Training Through Faster Data Access and Checkpointing DevFeed: [How Adobe Reduced GPU Idle Time in Generative AI Training Through Faster Data Access and Checkpointing]() Original publisher: [Read original article]() Author: Maddie Presland Published: 2026-08-25T15:22:30Z Content type: article Language: en Sources: [Backblaze]() Topics: [GPU](), [Training AI Models](), [cloud-infrastructure](), [Network](), [data]() Tags: [ai-ml](), [ai-training](), [b2cloud](), [cloud](), [cloud-storage](), [compute](), [ethernet](), [featured](), [featured-cloud-storage](), [gpu](), [model-training](), [networking]() ### AI overview Adobe's generative AI training pipeline left roughly two-thirds of GPU time waiting for data. The article attributes the waste to storage and retrieval bottlenecks, networking limits, and checkpointing overhead, and describes Adobe's use of a high-performance networking fabric and fragmented checkpoint storage to reduce delays. ### Source excerpt Adobe's experience reveals why GPUs sit idle during AI model training: slow storage, insufficient throughput, and uneven workloads. Learn how storage bottlenecks and uneven data loading leave expensive GPUs idle--and how always-hot, high-throughput object storage keeps AI training pipelines running efficiently at scale while reducing wasted compute costs. The post Why Your GPU Is Sitting Idle: The Data Pipeline Problem No One Talks About appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Results from the Backblaze Generative AI Media Hackathon DevFeed: [Results from the Backblaze Generative AI Media Hackathon]() Original publisher: [Read original article]() Author: Jeronimo De Leon Published: 2026-08-14T15:52:05Z Content type: article Language: en Sources: [Backblaze]() Topics: [Generative AI](), [Orchestration](), [cloud-infrastructure](), [For the Love of Code](), [Open Source](), [SDK]() Tags: [2026](), [ai-ml](), [b2cloud](), [cloud](), [cloud-storage](), [developer](), [featured](), [featured-cloud-storage](), [generative](), [hackathon](), [object-storage](), [open-source](), [orchestration](), [sdk](), [tech-lab](), [technical]() ### AI overview The article examines the strongest projects from the Backblaze Generative AI Media Hackathon and argues that production-ready generative media applications depend on more than generating images, video, or audio. It highlights storage design, provenance, verification, correction, deletion protection, and orchestration across multiple providers. ### Source excerpt See what the strongest projects from the Backblaze Generative AI Media Hackathon had in common. These five generative media apps show how storage and orchestration become part of production-ready design. The post Results from the Backblaze Generative AI Media Hackathon appeared first on Backblaze Blog | Cloud Storage & Cloud Backup ## Backblaze Drive Stats: How an Open Dataset Powers Academic and AI/ML Research DevFeed: [Backblaze Drive Stats: How an Open Dataset Powers Academic and AI/ML Research]() Original publisher: [Read original article]() Author: Stephanie Doyle Published: 2026-08-13T15:08:38Z Content type: article Language: en Sources: [Backblaze]() Topics: [dataset](), [NVIDIA DRIVE](), [Machine learning](), [Transformer](), [Disk image]() Tags: [ai-ml](), [articles](), [b2cloud](), [cloud-storage](), [data](), [drive](), [featured](), [featured-cloud-storage](), [hard-drive-stats](), [ml](), [research](), [source](), [storage]() ### AI overview The article explains how Backblaze Drive Stats evolved from an internal hard-drive reliability tool into an open dataset used in academic and AI/ML research. It describes the dataset's real-world scale, quarterly publication, SMART attributes, labeled failures, broad manufacturer coverage, and use in hard-drive failure prediction research. ### Source excerpt Backblaze Drive Stats has been cited in more than 105 academic papers and AI/ML projects since 2018. Explore the research it powers and download the dataset. The post Backblaze Drive Stats: How an Open Dataset Powers Academic and AI/ML Research appeared first on Backblaze Blog | Cloud Storage & Cloud Backup