회사 소식 Storage News Ticker - 18 September

Alation announced six new products for its enterprise AI operating system AIOS, including an industry-first tool for real-time enterprise AI agent compliance monitoring to resolve a key barrier to AI agent performance in large firms. AI Governance adds agent lineage tracing to track every agent’s regulatory risk down to live consumed data. Six native connectors integrate major AI agents and models into a cross-platform registry:

Amazon Bedrock, Amazon SageMaker, Databricks MLflow, Microsoft Copilot Studio, Microsoft Foundry, and Snowflake Cortex. It is the first cross-platform registry continuously linking agent compliance status to underlying data quality and policy state.

Console is a unified intelligent interface directing natural language requests to proper AIOS capabilities without manual navigation. AIOS enables intelligent apps to operate as one integrated system, as does its management OS.

Governed Collections makes non-catalog unstructured data such as compliance policies, SOPs and business glossaries available to agents as governed, up-to-date context. Documents from Confluence, SharePoint and S3 become governed catalog objects referenced by agents, which automatically load updated source files.

 Intelligent Feeds deliver governed interactive intelligence combining metrics, visuals, narratives and chat for recurring operational decision teams. Data intelligence shifts from static user visits to proactive insights with deep exploration support. Ontologies are new enterprise knowledge assets guiding agents precisely through business workflows.

Rather than letting agents infer rules, ontologies explicitly define governed, machine-readable business models mapped to catalogs and validated by domain experts. Built on open standards, they can be exported for any agent runtime.

Semantic Model Mastering imports semantic models from Databricks and Snowflake, enforces governance and enriches definitions, then syncs updates back automatically. Terms like “active customer” use one unified definition for all downstream agents and consumers, no manual re-entry required.

Chinese DRAM maker CXMT plans NAND chip production alongside DRAM.

Content platform Egnyte launched Egnyte Context Layer, mapping connections between content, people, projects, business systems and organizational knowledge. It pre-builds context maps for AI agents to cut redundant computation and costs while boosting result accuracy and speed. The layer powers task agents across financial services, life sciences, media, AEC and other vertical workflows.

Hitachi Vantara commits to global value chain net-zero emissions by FY2040; short and long-term targets are independently validated by SBTi.

• Cut absolute Scope 1 and 2 emissions by 98% by FY2030 (FY2024 baseline).
• Reduce Scope 3 emissions per usable petabyte of sold storage by 97% by FY2040. Hitachi Vantara cut Scope1/2 emissions by 43% in FY2025. Client efficiency gains include DestekBank’s 25% lower data center power use and Malayala Manorama’s 70% drop in power/cooling costs.

Tel Aviv, Sep 15, 2026 — Keewano released KeewanoDB, the first database built natively for scalable machine reasoning. The firm secured $12M seed funding led by Hetz Ventures with a16z speedrun, Remagine Ventures, DIG Ventures and angel investors participating. KeewanoDB records all user, device and AI agent actions in machine-optimized live, ordered, contextual form. In-database parallel analysis processes trillions of events and returns compact, context-grounded results for reliable agent reasoning.

Quantum appointed Mike Heuer as CIO to lead global IT covering enterprise applications, data, AI adoption, infrastructure and cybersecurity. Heuer previously served as CIO at Blackstone portfolio firm Legence, building its IT organization and supporting over 20 acquisitions. Earlier, as Blackhawk Network’s Global IT Head, he rolled out ERP and CRM systems across North America, Europe and Asia.

Redpanda together with Aiven, Confluent, StreamNative and Ververica formed the Streamhouse Working Group to create an open, vendor-neutral data architecture for AI agents and real-time applications. As businesses delegate more authority to software, agents cannot rely on stale data and need continuously current governed business views.

Streamhouse architecture delivers production-grade real-time business state for apps and agents without full enterprise data consolidation. The group will maintain the open definition and advance related technologies, standards and interoperability, ensuring no single vendor controls this architecture.

Storage virtualizer StorMagic and Mako Networks partnered to build more resilient, manageable remote/distributed IT environments. The joint solution pairs StorMagic SvHCI edge hyperconverged infrastructure with Mako’s PCI-certified cloud-managed SD-WAN and networking platform for centralized management, secure connectivity and failover. Combined edge networking and compute deliver tangible operational benefits for modern retail.

StorONE announced workforce solutions provider Engage2Excel expanded its StorONE deployment to cover production storage, backup, VTL, business continuity, disaster recovery and long-term retention. Engage2Excel manages over one petabyte of enterprise data on StorONE systems. StorONE FlashSpan technology improves usable flash capacity up to 9x by dynamically analyzing access patterns and placing data optimally to reserve flash for high-value workloads.

HDD vendor Western Digital highlights AI’s next infrastructure challenge: storing exploding persistent datasets. IDC surveyed over 700 AI infrastructure and storage decision-makers:

• 95% store more data due to AI
• 95% report rising data value
• Nearly 75% retain data longer
• Over 75% reactivate archived data for AI workloads, blurring cold and active data boundaries. Growing data generation, retention and reactivation make storage capacity, accessibility and cost critical infrastructure priorities.

Yugabyte recorded its strongest first half ever: paid new customers rose over 125% YoY, and more than 900 companies downloaded YugabyteDB. Its May-launched agent-native context engine Mekodata.ai gained early traction, with agent-created database volume growing 10x daily. Yugabyte also added key leaders across engineering and go-to-market including EMEA VP Sales Sonny Purewal and India GTM VP Linson Mathai. Expanded product lines and partner programs support enterprise AI and agent application rollouts.

SW RAID vendor Xinnor published Weizmann Institute PoC results for a new Tier-0 storage design for HPC and AI. xiRAID Opus converts existing NVMe drives inside GPU and compute nodes into shared protected storage pools. Modern GPU servers have local NVMe typically limited to single-node scratch use; xiRAID Opus disaggregates these drives across clusters into software-defined shared storage with RAID protection, multipathing and failover.

The PoC stacked IBM Storage Scale atop xiRAID Opus block storage to deliver shared file systems and global namespace. The system hit 131.7 GiB/s sequential read, nearly matching raw NVMe-oF speed, and 46.9 GiB/s sequential write retaining ~98.5% raw bandwidth. Random 4KiB IO reached 5.8M read IOPS and 3.1M write IOPS. With Nvidia GPUDirect Storage, aggregate reads from shared storage into GPU memory across five nodes hit 132.1 GiB/s.

Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!