# Versity says tape libraries can be a data source for GPUs doing AI work > Source: > Published: 2026-08-04 16:00:00+00:00 # Versity says tape libraries can be a data source for GPUs doing AI work You might have to read this twice; [Versity](https://www.blocksandfiles.com/data-management/2023/06/13/versity-open-sources-s3-gateway-with-no-poison-pill/1610800) has added a high speed S3 over RDMA capability to its ScoutAM archive gateway software, meaning tape archives can now pump data directly into a GPU’s high-bandwidth memory. The open-source Versity gateway supports S3 over RDMA (Remote Direct Memory Access) through Nvidia’s cuObject framework. RDMA accelerates data transfer between the archive’s S3 gateway and the S3 client. Behind that fast front door, Scale Out Archive Manager (ScoutAM) manages the tape library, writing data to tape and reading it back at scale. The result is a GPUDirect Tape Gateway that pairs full-speed RDMA at the front end with the capacity and durability of tape archive on the back end. Versity CEO and co-founder Bruce Gilpin tells us: “With the cost of NVMe and memory skyrocketing and a 2U server now back-ordered for months, people are getting desperate to build out AI infrastructure with unconventional and unconstrained storage systems. Tape has not been used much in AI because the players were financially unconstrained. But that is changing, partly due to budget issues but more due to availability. “Tape drives stream data faster than people realize and you can run them in parallel. 50 drives can generate a 20 GBps stream to feed GPU’s and it’s dirt cheap. One big player is experimenting and I think we will see others. Lucky for us, you need a super capable software stack to orchestrate all of this but the base storage system is a bunch of cheap tapes and relatively cheap drives and libraries.” The Versity Gateway is a stateless, modular, open source S3 gateway that translates standard AWS S3 commands into operations against a range of backend storage systems, including POSIX filesystems, [ScoutFS](https://www.versity.com/products/scoutfs/)/ScoutAM (supporting tape), and cloud storage. The new architecture adds an RDMA-based data path alongside the gateway’s existing TCP control path. It scales by distributing many GPU or host client systems across any number of gateway server instances, each pair establishing its own direct RDMA path between client memory and gateway memory. Versity says RDMA allows data to move directly between the memory of one system and the memory of another, without routing through the host CPU. NVIDIA’s cuObject framework, part of the GPUDirect Storage ecosystem, applies this to S3 compatible object storage: it gives applications a way to move data directly between GPU or system memory and object storage services using RDMA transports such as InfiniBand or RoCE. It explains that cuObject splits an S3 request into two separate channels. Session setup, authentication, bucket verification, and object metadata flow in-band, through the standard S3 API, exactly as they would for any other request. The object data itself moves out-of-band, over an RDMA connection that runs alongside the S3 session rather than through it. The cuObjClient manages this data flow on the requesting side and the cuObjServer manages it at scale on the gateway side. The RDMA data channel never passes through the client’s CPU and never passes through the S3 protocol stack, so it avoids the overhead that the protocol would otherwise add to the transfer. For workloads that involve GPUs specifically, data can travel directly between the network card and GPU memory, skipping the CPU almost entirely. A GPU host system — GPU, CPU, NIC — connects over RDMA (RoCE/IB) directly to a Versity Gateway (VGW) server system’s NIC and into VGW memory. Backend storage sits behind the gateway and is reached through the gateway’s normal path, not RDMA. Any number of GPU host clients and any number of VGW server instances can be deployed this way.) For a PUT request, the object is uploaded to the gateway’s memory via RDMA, which the gateway then streams to the backend storage. For a GET request, the data is first staged into the gateway’s memory from the backend storage and then transferred to the client over RDMA. This enables client RDMA access to any of the gateway’s supported backends. ScoutAM’s GPUDirect Tape Gateway addresses a problem that has become more pressing as datasets grow: how to work with enormous volumes of data on cost-effective, durable media without turning storage operations into a slow-down drag on GPU-driven pipelines. Archiving data to tape is likely to remain what it has always been: a steady, ongoing ingest process. Where RDMA changes the equation is on the way back out. AI training pipelines don’t read a dataset once; they reread it every time a model needs to be retrained or a dataset needs to be reprocessed. ScoutAM can reorder large data set requests to match the optimal tape offset request locations. It can minimize tape mounts, and maximize streaming data performance. Combined with the new capability to utilize RDMA transfer to GPU, recalling massive datasets from tape becomes a feasible approach for model training and other large scale data processing pipelines. Versity says organizations running AI training pipelines, large-scale machine learning jobs, or HPC applications often find their storage layer, not their compute layer, is the limiting factor. Because the performance gain is most pronounced on the read side, the workloads that benefit most immediately are the ones defined by repeated, large-scale reads: loading and reloading training data into GPU memory, feeding checkpoints back into resumed training runs, and rereading existing datasets for new analysis. For tape specifically, this changes what’s practical to keep in an active archive rather than treating it as cold and slow storage. Training datasets, checkpoints, and simulation outputs that might previously have felt too slow to recall efficiently can now be pulled back at RDMA speed when it’s time to use them again. Organizations in fields such as genomics, media and entertainment, scientific research, and large-scale AI development, all of which routinely generate data volumes measured in petabytes, get a faster path both to long-term retention and back out again. Data still arrives and is retrieved as standard S3 objects; POSIX filesystems and ScoutFS continue to work as they always have, and ScoutAM still manages tape the way it always has. Because the Versity Gateway is open source and includes RDMA support as part of the same project rather than as a separate paid add-on, teams can add this capability to existing S3 workflows without adopting a new proprietary gateway or a new license. MinIO’s [AIStor](https://www.blocksandfiles.com/ai-ml/2026/07/30/minios-aistor-memory-enables-agents-to-inherit-organizational-memory/5281348), which was announced in late 2024, also supports fast S3. It extended MinIO’s Enterprise Object Store software with the S3 API, PromptObject, and support for S3 over RDMA, to provide fast object storage for AI training and inferencing. MinIO says data moves directly from AIStor to GPU memory (HBM) over RDMA, bypassing the CPU and system/host memory entirely (zero-copy, kernel-bypass). However, it is a paid and proprietary product unlike Versity’s open-source GPUDirect Tape Gateway, which ships as part of the Versity S3 Gateway. RDMA and cuObject support, including the GPUDirect Tape Gateway integration with ScoutAM, are not a separate paid tier. Bootnote In [June ](https://www.blocksandfiles.com/file/2026/06/12/beegfs-and-grau-data-add-tape-archive-backend-to-parallel-file-system/5254919)BeeGFS parallel filesystem supplier [ThinkParQ](https://www.blocksandfiles.com/nvme/2019/05/28/excelero-boosts-beegfs-for-faster-ai-work-by-gpus/1613296) and tape archive supplier [GRAU DATA](https://www.blocksandfiles.com/ai-ml/2024/12/11/panzura-makes-symphony-data-intelligence-music-with-graus-metadata-hub/1588106) set up a strategic partnership through which the latter's XtreemStor tape archive system has been integrated with BeeGFS. This, the two say, provides a storage infrastructure that can handle the high performance of data-intensive AI, HPC and big data workloads and the long-term storage of the data needed for, and generated by, them.