Executive summary
- Hitachi Vantara announced several new enhancements to its VSP One Object storage platform, including an increase of up to 5.5x in performance for small objects and support for querying open-format data directly on the platform.
- The announcements align to an industry-wide transformation of data storage capabilities, spanning block, file and object, to better serve AI environments.
- The increased demands of AI, especially with the increased use of agents, necessitate data storage modernization to ensure infrastructure environments are fully optimized to maximize ROI.
Last week, Hitachi Vantara announced multiple enhancements to its VSP One storage platform to better support enterprise AI initiatives. These announcements join a larger industry-wide movement to transform enterprise storage to better serve the specific demands of AI environments.
What did Hitachi Vantara announce, and why is it important?
With its VSP One Object expansion, Hitachi Vantara announced the following new features:
- A smaller, four-node starting configuration for VSP One Object, with enhancements providing an increase in performance up to 5.5x for small objects.
- Improved multisite and regional replication capabilities.
- Support for querying open-format data, including Iceberg, Parquet, JSON and CSV, directly on the VSP One Object platform.
Architecting an end-to-end AI storage ecosystem spanning data preparation, model training, tuning and inference necessitates a diverse mix of storage technologies, often demanding a multiprotocol approach that spans object, file and block storage. While innovation is underway across nearly every type of data storage technology to better support AI, these three Hitachi Vantara announcements align with a larger, industry-wide transformation within object storage, fueled by demand for faster and more frequent data access.
The appeal for object storage for AI stems from its architecture, leveraging a flat address space designed to support massive, multi-petabyte stores of unstructured data while providing high levels of resilience. This design has led to its use in data lake environments, supporting large volumes of text, video, image and audio data. Hitachi Vantara's addition of support for open-format queries helps simplify the process of managing the data sets stored within VSP One Object, which in turn helps accelerate time to value for larger AI projects.
Object storage's use in data lake environments suits its role in AI data preparation, while its ability to provide cost-effective capacity helps it serve as a storage target for model checkpointing. The architecture of traditional object storage, however, tends to be better optimized for sequential-access, read-heavy environments rather than high-frequency, random data read/write access patterns.
Therefore, it is not surprising that research from Omdia, a division of Informa TechTarget, revealed the following:
- 71% of IT decision-makers agreed that effectively integrating storage with data pipelines and AI workflows is a major challenge.
- 70% agreed that storage-related challenges are a significant barrier to AI success.
The increased performance demands of AI initiatives have transformed what businesses require from object storage. In the AI era, object storage platforms can no longer simply serve cold archives of data. Object storage must evolve to provide increased transactional performance, ensuring that the increased demand for data does not impede time to value for AI operations. This demand for greater performance is driving Hitachi Vantara's announcement.
With agentic AI on the rise, data access will become even more frequent and less predictable, driving up demand for performance across block, file and object storage. Storage decision-makers must raise their expectations of storage providers, especially when it comes to performance, if they want to maximize the ROI from their AI investments.
Can I use my existing storage environment for AI?
Based on the diverse and extensive demands of enterprise AI, your existing legacy storage environment will not be fully optimized to ensure your organization can get the maximum return on its AI initiative investments. According to the previously mentioned Omdia research, 83% of organizations said they expect to upgrade their storage environment over the next 24 months to support their growing AI demands.
Early on in your AI journey, your organization will likely be able to make do with a mix of existing and new storage infrastructure. The optimal infrastructure ecosystem, however, will depend on your specific AI use cases and approach. The storage you need for training your own unique models is different than the storage environment you need for tuning an off-the-shelf model or for RAG.
The initial priority for AI storage investments should be to ensure that your GPU or accelerator investments are fully utilized. After that, subsequent priorities likely will include improving the AI experience in production with added performance, or accelerating data preparation activities to ensure that data sets are in the best shape to support your specific AI needs. Similar to Hitachi Vantara's support for querying open-format data, other data storage offerings are increasingly adding new capabilities to support data management activities. In the AI era, data storage is not just storing and serving data; it is helping manage it as well.
Given the diverse and extensive demands that enterprise AI environments are placing on data and data storage ecosystems — especially with the increase in agentic adoption — your existing storage ecosystem will have to evolve to support your organization's AI objectives.
Ultimately, organizations must take in the following considerations:
- AI is not just another enterprise workload. It is a category of diverse use cases and approaches that increase the performance, scale and governance demands placed on data storage while increasing complexity and reducing predictability when compared with legacy application environments.
- Achieving positive ROI on AI investments necessitates ensuring that the infrastructure is optimized to the needs of your specific AI environment. Ensuring constant optimization of the entire data path, including GPU investments, at scale necessitates innovation and increased data management capabilities from the data storage layer.
- These innovations create a new and rapidly evolving era of storage differentiation, one that storage decision-makers must incorporate in their decision-making process.
Nearly every data storage vendor, including Dell Technologies, Everpure, HPE, Hitachi Vantara, NetApp, Vast Data and Weka, is evolving its portfolio to address the specific needs of AI. The bottom line is that your storage must evolve for AI, and the industry has already taken note.
Scott Sinclair is Practice Director with Omdia covering the storage industry.
Omdia is a division of Informa TechTarget. Its analysts have business relationships with technology vendors.