Generative AI (GenAI) is taking the world by storm. Its potential to revolutionize how we engage with our data, with our customers, and even our infrastructure is being quickly realized by organizations savvy enough to embrace the technology. At the same, GenAI is complex, and doesn’t look like anything we’ve seen in a data center before.
Dell Technologies announced a significant expansion of its Generative AI Solutions portfolio to empower businesses embarking on their GenAI journeys. This expansion delivers advanced infrastructure, professional services, and a collaborative data solution to help organizations derive intelligence from their data securely and efficiently.
New Validated Design for Generative AI
Dell has long taken a validated design approach to simplifying complex solutions. Examples of this are Dell’s validated solutions for VDI, HPC, Analytics, AI, and a variety of high-performance software-centric configurations. Dell is taking the same approach with generative AI, announcing its new Validated Design for Generative AI in collaboration with NVIDIA.
The new validated design for GenAI is a notable addition to the portfolio, offering optimized hardware solutions and pre-trained models that allow users to extract insights from data. It also gives customers a path to customize GenAI models according to their unique objectives – an approach that streamlines the development of GenAI models while ensuring data security.
AI Models
Maximizing the value of large language models (LLMs) requires organizations to combine and train models on their data, aligning with specific use cases. A robust data management strategy is essential to ensure data quality then customize pre-existing LLMs to integrate domain-specific information.
Tuning GenAI models with domain and organization-specific data is a critical step. Rather than training models from scratch, models are supplemented with data relevant to the application. Standard customization techniques include transfer learning, fine-tuning, instruction tuning, prompt learning, and reinforcement learning.
The Dell Validated Design for Generative AI now supports model customization, tuning, and inferencing. This dramatically simplifies GenAI model implementation.
Built on PowerEdge & PowerScale
The architecture is documented for Dell PowerEdge XE9680 and PowerEdge XE8640 servers, including a choice of NVIDIA Tensor Core GPUs, coupled with NVIDIA AI Enterprise software, NVIDIA NeMo™ framework, and Dell-provided software.
Compute is combined with storage options like Dell PowerScale F600 and Dell ObjectScale, offering scalability for operationalizing GenAI models and accommodating various data types.
You can roll your own using the validated design, but the Dell infrastructure components are also available with flexible consumption models through Dell APEX. Using Dell APEX makes a lot of sense for organizations that are just starting their AI journey, allowing the flexibility to build and tune the infrastructure over time.
The new validated design will be available later this month.
New Data Lakehouse for AI
Many enterprises struggle to harness AI’s full potential due to difficulty of locating, accessing, and processing data across disparate environments. The truth is that most data remains on-premises, forcing organizations to choose between complex tool integration or replicating data to the public cloud.
In collaboration with Starburst, Dell Technologies addresses these challenges with a new, open, modern data lakehouse. The offering includes a highly secure single access point for all enterprise data, interoperability with on-prem and cloud tools, a unified stack for various workloads, high performance with decoupled compute and storage, predictable costs, a simplified user experience, and reduced data movement.
The new solution combines Starburst’s analytics software with Dell’s PowerEdge compute and storage platforms, enabling customers to establish a high-performance lakehouse on-premises and in multi-cloud environments. This offers a cost-effective alternative to cloud-based options and faster ROI by simplifying deployment and tapping into existing datasets.
The solution is built on the open-source, high-performance Trino distributed SQL engine, known for its speedy analytic queries against data lakes, lakehouses, and distributed data sources. By combining Starburst with Dell’s compute and storage platforms, the companies aim to deliver a cost-effective, scalable, on-premises, multi-cloud lakehouse solution for modern organizations’ analytics and AI workloads.
The collaboration will deliver a cost-effective and scalable lakehouse solution that leverages open table formats like Iceberg and Delta Lake. The solution will be available in the first half of 2024.
New GenAI Professional Services
Dell also added to its GenAI professional services portfolio to better support customers in preparing their data and creating operational GenAI platforms. This includes services for data cleansing, integration, and quality assurance.
One of the most interesting new offerings is the new Data Preparation services. Data preparation services allow organizations to ready their data for AI usage. You can specify the data requirements for GenAI, establish data pipelines to categorize, clean, label, and anonymize the data, and populate databases like LLM databases, such as vector databases.
Dell also introduced its Managed Services for Generative AI for customers who like a managed infrastructure. This allows customers to focus on delivering AI-enabled solutions to their organization while letting Dell manage the details of the AI infrastructure.
There are also new training options. Dell Education Services for Generative AI offers a wide range of training courses to address IT organizations’ AI skills shortage, covering topics like data engineering, large language model deployment and customization, and NVIDIA hardware and software administration. This includes a new Accelerator Workshop, provided at no cost, to help organizations speed up implementing their GenAI strategies.
The new service offerings go a long way towards expedited AI project execution while providing organizations with the essential skills needed for success.
Analyst’s Take
Dell takes the potential of AI seriously. Just last week, Dell COO Jeff Clarke announced that Jeff Boudreau will fill the newly created position of Chief AI Officer. In that role, Boudreau and his team will partner across the company to understand domain-specific use cases, build, define, and standardize architectures for the future, and integrate AI across Dell’s entire product portfolio. The team will also build relevant AI partnerships, lead the Center for AI Innovation that will set Dell’s policies, and help educate its team members.
Generative AI is complicated and arrives with a steep learning curve. Dell removes much of the complexity with the expansion of its Generative AI Solutions portfolio. The new offerings underscore Dell’s commitment to facilitating its customers as they embark on their AI journeys.
The introduction of validated designs, professional services, and innovative data solutions will enable organizations to maximize the potential of AI and analytics while accelerating time-to-value for GenAI projects. Making the validated designs available on Dell APEX offers a nice degree of flexibility. All of this removes the burden from IT practitioners while accelerating time-to-value for AI projects.
As the number one infrastructure company on the planet, Dell Technologies is ideally positioned to help enterprises tackle the complexities of AI. Today’s announcements demonstrate that Dell takes that position seriously.
Disclosure: Steve McDowell is an industry analyst, and NAND Research an industry analyst firm, that engages in, or has engaged in, research, analysis, and advisory services with many technology companies, which may include those mentioned in this article. Mr. McDowell does not hold any equity positions with any company mentioned in this article.
Read the full article here
