Infinidat G4 News Focuses on Enabling Power and Efficiency for Enterprise AI Applications
By Jean S. Bozman, Cloud Architects LLC
Infindat announced a set of new innovative solutions. Here is Jean Bozman’s take on these announcements. As AI rapidly transforms customers’ enterprise IT landscapes, AI workloads and AI-enabled applications are growing worldwide. This is a dynamic force that is expanding the total amount of enterprise data that must be aggregated, stored, and managed in customers’ data centers and remote locations.
Infinidat, based in Waltham, Mass., and Herzliya, Israel, announced significant capabilities that bring high-capacity storage in smaller “footprints.” With its September announcements for the firm’s G4 storage system, Infinidat is bringing compact, energy-efficient high-end solutions, while it is adding lower entry price points, as well.
Providing a wider set of price/points and capabilities will only gain in importance, especially with the pending acquisition by worldwide provider Lenovo, as announced in January 2025, which is subject to customary regulatory approvals.
This expanded range of Infinidat’s InfiniBox G4 family of all-flash storage products – already known for their 100% availability guarantee and cyber-resiliency – will fit a broader range of use-cases for customers’ AI infrastructure and enterprise workloads.
Providing a wider set of price/points and capabilities will only gain in importance, especially with the pending Lenovo acquisition announced in January 2025, which is subject to customary regulatory approvals.
Key Points
Customers are finding that rapidly growing AI deployments are straining the capacity and budgetary limits of their on-premises data centers.
By providing more capacity in a smaller footprint, Infinidat is addressing customer needs to reduce power/cooling in highly dense data-center racks, while keeping CAPEX and OPEX costs in check.
The Infinidat Infinibox G4 news is important for several reasons, because it:
- Delivers a high-end, 100% availability enterprise storage solution in a reduced footprint. The new InfiniBox SSA G4 F24 all-flash models deliver high-end enterprise storage functionality in a small form-factor of only 11RU, delivering up to two times (2X) performance of the InfiniBox SSA G4 – and delivering it with enhanced Green IT power-efficiency as well. The 11RU InfiniBox SSA G4 F24 starts at an entry capacity of 77 TB.
- Increases capacity for the InfiniBox hybrid platform’s largest model up to 33 petabytes (PB) effective capacity – far outstripping the previous 17.2 petabytes (PB) effective capacity for InfiniBox hybrid systems announced in May 2024. We note here that the 33PB-effective InfiniBox hybrid solution will be a powerful backup target device for the enterprise – providing shorter backup windows, reduced backup restoration times, and cyber-storage resilience and recovery.
- Protects RTO for enterprise mission-critical workloads. The Infinidat InfiniBox SSA G4 systems ships with a guaranteed Recovery Time Objective (RTO) of one minute or less, regardless of the dataset size.
- Supports multiple data types natively in Infinidat’s InfuzeOS storage operating system throughout the enterprise. Customer data comes from multiple sources – including block, file and S3 object data-stores, including data from VMware and Red Hat OpenShift data sources. This broad approach reflects most enterprise customers’ mix of data formats (file, block, and object). Without that broad native storage OS support for data-types, customers might find that their enterprise data models could become fragmented, producing less reliable AI results.
- Improves power efficiency for a broad range of customers, including data that is being generated in enterprise customers’ data centers, remote facilities (e.g., factory data centers and distribution-site data centers), and in customers’ extended hybrid multi-cloud environments. Overall power consumption is becoming a key factor for enterprise decisions to build power-hungry data centers worldwide.
- Reduces the entry price point for more powerful enterprise storage configurations, bringing customers a new NVMe TLC flash offering.
- Adds capabilities and improved price/performance for InfiniBox SSA G4 F24 all-flash systems, which are often sold through a worldwide channel ecosystem of Infinidat’s solution partners. The G4 series of systems was introduced last year, entering a dynamic storage market in which Dell Technologies, NetApp, Pure Storage, HPE, IBM, and others compete across a series of price ranges.
Announcement Details
This year’s Infinidat G4 offers are focused on mission-critical levels, acknowledging the increasing density and capacity requirements for enterprise-level scale-up AI and other critical enterprise applications and workloads.
Price performance is improved, with configurations that are 30% smaller than earlier models – and which are offered at lower price-points than before. The new systems require 45% less power per petabyte (PBu) than for earlier G4 models. Focusing on AI-driven enterprise capabilities enhances the company’s continuing competition with a range of enterprise storage providers, such as Dell Technologies, HPE, and NetApp, and others.
We know that customer data comes from multiple sources – including block, file and object data stores. This reflects most enterprise customers’ mix of data formats (file, block and object). However, in many places, these data-types are managed and maintained in “silos” as separate data-types.
Importantly, the Infinidat systems support all these major data-types natively in the InfuzeOS storage operating system that powers all InfiniBox models. This addresses customer deployments for AI training, AI modelling, and AI inference scenarios. That approach includes a broader portfolio of enterprise data – with the aim of making AI data analyses more inclusive and more accurate.
The Infinidat G4 storage system’s software profile includes an enhanced InfuzeOS operating system and a full stack of InfiniSafe software for comprehensive cybersecurity that is included at no additional charge.
Software management is key to including a variety of data types (e.g., VMware, OpenStack, and traditional file systems) – along with unstructured data in AI workloads. The InfuzeOS supports embedded native software stacks – adding support for S3 Object data-types used in cloud and hybrid-cloud infrastructure to the native existing block and file support.
Key Takeaways
AI infrastructure – and scale-up AI in particular – are fast becoming the primary focus for enterprise data centers, CSPs, and even hyperscalers (e.g., AWS, Microsoft Azure, and Google Cloud). supporting enterprises.
Enterprise requirements are forcing a new wave of designing, building, and protecting scalable AI infrastructure, making sure to include a wide spectrum of data and data-types.
All major system vendors and storage vendors are feeling that pressure to include a range of enterprise data for AI workloads – along with the need to keep power/cooling, CAPEX and OPEX costs in-check for enterprise customers.
Enterprise Capabilities
Enterprise-level “ilities” are growing increasingly important to enterprises that realize they must protect their AI-enabled applications – by ensuring their availability, cyber-security and power-efficiency. We note here that Infinidat’s key values of ensuring enterprise-level cyber storage security and guaranteed high-availability are important attributes that are vital to enterprise customers’ use cases.
Systems providers must balance the “need for speed” with practical considerations about acquisition costs and ongoing maintenance costs. Without that balance, enterprises might reduce their own on-premises installations in favor of tapping IT and cloud services from outside providers.
As the AI infrastructure boom continues, most customers will likely adopt a mix of both: hosting AI workloads running in their on-premises data centers as much as possible, while leveraging cloud services from cloud-service providers, depending on the types of AI enterprise use-cases for their business units worldwide.
