Virtualization Technology News and Information
VMblog Expert Interview with Lightbits Labs: A "Block" Solid Foundation for AI Cloud Services and Data Pipelines

interview lightbits gordon 

As organizations strive to unlock the full potential of artificial intelligence (AI) and cloud-native application development, they face a multitude of challenges. From capacity planning and resource utilization to balancing performance and cost, the journey towards an efficient AI cloud is fraught with obstacles. Enter Lightbits Labs, a company at the forefront of revolutionizing block storage solutions for AI workloads and data pipelines.

In this exclusive Q&A session, we sit down with Abel Gordon, Chief Technology Officer at Lightbits Labs, to explore the pivotal role of block storage in accelerating AI pipelines and maximizing resource efficiency. Gordon sheds light on how Lightbits' cutting-edge technology can streamline data pre-processing, minimize latency for real-time decision-making, and deliver enterprise-grade resiliency – all while ensuring unparalleled performance and cost-effectiveness.

With a deep understanding of the intricate demands of AI workloads, Gordon provides valuable insights into the unique capabilities of Lightbits' software-defined storage solution. From proactive node evacuation to flexible scalability, he illuminates how Lightbits empowers organizations to tailor their storage infrastructure to meet the specific requirements of diverse AI workloads, ultimately driving innovation and competitive advantage.

VMblog:  How can organizations leverage block storage to build a faster, and more cost-efficient AI cloud?

Abel Gordon:  When it comes to architecting a data platform for AI workloads and cloud-native application development, enterprises need to overcome several challenges. These include everything from capacity planning and balancing performance and cost to maximizing the utilization of resources.

AI clouds demand exceptional performance and efficiency at scale. While large datasets typically reside in a data lake optimized for capacity efficiency, various critical stages of AI pipelines demand high-speed and low-latency storage. This is where organizations have found a need for fast, scalable block storage architected for NVMe-playing a key role in accelerating AI pipeline performance and maximizing resource efficiency.

At Lightbits, AI cloud providers, like Crusoe, use our fast block storage solution in several ways. With Lightbits, they get clustered storage with performance equivalent to local flash, but with much better resource utilization, plus essential data services such as snapshots, clones, replication,  and compression.

VMblog:  So, how does Lightbits fit into the AI data pipeline and how can it improve AI workload performance?

Gordon: Lightbits can accelerate AI workflows in several ways. First, it can streamline data pre-processing (the phase where raw data is cleaned and transformed into a format that can be effectively used in machine learning models) at the ingest stage. For instance, an image processing pipeline that normalizes and augments millions of images for a deep learning model can perform these tasks with higher efficiency when the intermediate data and tuning parameters are stored on Lightbits instead of local NVMe SSDs or other types of storage. So, by using Lightbits technology to store checkpoints, raw data, parameters, labels, and so on, you will boost your infrastructure, accelerate processing, reduce your tail latency, and increase the overall throughput.

Another way Lightbits can fit into the AI pipeline and improve workload performance is by ensuring minimal latency to query information from vector databases, and other databases like MongoDB, Elasticsearch, and PostgreSQL. This is extremely important for applications where decisions need to be made in fractions of a second. Just think of the positive impact this has on e-commerce and financial services customers who need to make instant decisions for fraud detection, personalization, and other real-time decisions.

VMblog:  What makes Lightbits a good choice to enable AI workloads?

Gordon: For AI workloads, Lightbits can deliver 16x higher performance block storage and consistently low latency with fewer servers compared to competitive software-defined block storage offerings. Lightbits also provides enterprise resiliency with more replication options.

The Lightbits software-defined storage solution provides flexibility and scalability, allowing customers to tailor their storage infrastructure to meet the specific requirements of AI workloads. Organizations with diverse AI workloads and a need for flexible, high-performance storage can significantly benefit from Lightbits' software-defined storage solution.

Take the Lightbits proactive node evacuation feature as an example of maintaining system integrity for AI instances. This feature is designed to safeguard data during server maintenance or when disabling operations. With user-triggered data eviction, all resources associated with a server, including volumes and snapshots, are seamlessly migrated to other servers within the cluster. This proactive approach ensures that volumes and snapshot data are fully protected before any maintenance operation occurs. For organizations building an AI Cloud service, this is a powerful feature that provides greater control over platform management and system maintenance processes.

VMblog:  Is Lightbits block storage only used for primary storage in an AI cloud?

Gordon: Our customers use Lightbits for AI in several ways. For example, our AI cloud services provider customers are storing and managing their customers' custom operating systems images using Lightbits-managed storage. Others have selected Lightbits for their vector databases as well as for MongoDB, Elasticsearch, other NoSQL databases, PostgreSQL, and more.

At its core, Lightbits is a high-performance block storage solution that enables users to disaggregate their NVMe storage to improve cost efficiency and resilience. Any file system requires an underlying block storage. This means customers can deploy their preferred file system on top of Lightbits and apply virtually any use case to benefit from the scale, performance, and efficiency boosts provided by our technology.

VMblog:  What are some of the other common use cases for Lightbits?

Gordon: There are a myriad of use cases for Lightbits. Many organizations build a cloud service using Lightbits, replacing Direct Attached Storage (DAS) and iSCSI architectures that were chosen to deliver high performance. With Lightbits, they get a disaggregated storage solution that can increase their utilization and improve resiliency while retaining high performance.

Others are using Lightbits to replace fibre channel SAN storage and migrate their virtualized workloads to Azure while retaining all the benefits of a SAN architecture. With Lightbits for Azure VMware Solution, they get the performance and data services of a SAN but with better resiliency, availability, and cost-efficiency This ultimately reduces their cloud storage TCO.

Furthermore, Lightbits has been found to dramatically improve operational efficiency for containerized applications with latencies that are as good or better than DAS. All this is done with simple and efficient NVMe over TCP. There's no need to touch the clients or the network, containers are easy to deploy at scale with Lightbits. As container ecosystems mature, constraints on the implementation of containerized applications keep falling away.

Published Wednesday, April 10, 2024 9:30 AM by David Marshall
Filed under: ,
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<April 2024>