HPE Alletra X10000 & SingleStore for AI Workloads

Name: HPE Alletra X10000 & SingleStore for AI Workloads
Uploaded: 2026-09-01T08:33:49-04:00
Duration: 4 min 15 s
Description: Explore the possibilities of creating a highly scalable, highly performant AI development environment by deploying the HPE Alletra Storage MP X10000 object storage system as the unlimited storage host of a SingleStore distributed database platform. See...

Hewlett Packard Enterprise

09/01/2026

0 (0%)

Report Like Favorite

Transcript

maximize the utility of their gathered data sets, but simply wanting to incorporate these elements and actually doing them are two very different things. Quality AI deployments require access to enormous reams of data and answers from that data as quickly as possible. The IT departments managing these environments need the AI dev architecture to be as easy to manage as possible without blowing up the budget. Starting with one's existing infrastructure is a good way to keep costs down, but usually that means not being able to scale effectively, dealing with storage platforms that can grow with the data but can't match the needed speed and complicated database setups. The answer lies in deploying a solution developed with the HPE Electra Storage MPx10,000 object storage system combined with the single store distributed database platform. The X10,000 allows IT staff to grow the compute and capacity for their S3 buckets independently to pinpoint exactly where help is needed. Single stores hybrid transactional and analytical processing database or HTAP allows multiple types of data to be managed under one database umbrella. The X10,000 and single store work together to simplify data management and provide a powerful and expertly scalable AI development solution. So let's go into a bit more detail on each of these products. The single store HTAP philosophy groups transactional data into row stores while analytics and unstructured data land in column stores. This architecture allows queries to return faster while simplifying data management in one product. The single store cluster structure makes it easy to scale by simply adding nodes. Single store automatically arranges the data across the cluster to use the new node count and it can handle data analytics itself without needing additional plugin apps to provide that capability. The X10,000 fits in this solution by being the repository for single store unlimited storage which is designed to house very large data sets on S3 compatible buckets. The local nodes of the cluster maintain the hot data that's frequently queried but if need be the queries can be directed to the unlimited storage bucket. In our solution column stores are configured to house the data on the X10,000 S3 buckets where they can take advantage of the X10,000's all NVMe storage that lowers the latency for queries of cooler data. This helps single store by ensuring that the cluster doesn't need to scale as quickly to make room for more hot data storage just to keep up the query speeds. To sum up, a deployment of single store using the X10,000 to house unstructured column stores means that AI developers get faster access to the latest data and at the performance needed. Customers with sensitive data can keep it in house on the X10,000 where they're in control and still have the option to pay as you go just like the cloud. IT staff can limit hardware growth to only what's necessary and have less to manage and get the simplicity of the HPE GreenLake platform to allow X10,000 deployment and administration from anywhere. Both now and going forward, an AI solution with single store and the HPE Electra Storage MPX10,000 will provide a high performance AI development environment. For more information on the X10,000 single store AI solution and some very compelling test results using TPCC and TPC-H datasets, check out the blog post at the Around the Storage blog page as well as the X10,000 and single store web pages. Thanks very much for watching.

TL;DR

AI workloads require massive datasets and fast query response, but most existing infrastructure cannot scale storage and compute independently without significant cost and complexity.
SingleStore's HTAP architecture unifies transactional and analytical data under one database, eliminating the need for additional plugin tools and simplifying data management.
The HPE Alletra X10000 serves as SingleStore's unlimited storage backend via S3-compatible buckets, using all-NVMe storage to reduce latency for queries against cooler, less-frequently accessed data.

Summary

This explainer video outlines how combining the HPE Alletra Storage MP X10000 object storage system with the SingleStore distributed database platform creates a high-performance, scalable AI development environment. The video opens by framing the core challenge facing enterprise IT teams: AI workloads demand access to massive datasets with low-latency query response, yet most existing infrastructure cannot scale storage and compute independently, and traditional database setups add operational complexity. The proposed solution pairs SingleStore's Hybrid Transactional and Analytical Processing (HTAP) architecture — which organizes transactional data in row stores and analytics or unstructured data in column stores — with the X10000's all-NVMe, S3-compatible object storage. Column stores are configured to reside on X10000 S3 buckets, keeping hot data on local cluster nodes while cooler data is served from the X10000 with low latency, reducing pressure on the cluster to scale prematurely. The X10000 also supports independent scaling of compute and capacity, and the HPE GreenLake platform enables pay-as-you-go consumption and remote administration. The video closes by directing viewers to benchmark results using TPC-C and TPC-H datasets available on the HPE Around the Storage blog.

Chapters

0:00 - The AI Infrastructure Challenge
0:56 - Introducing the Joint Solution
1:41 - SingleStore HTAP Architecture
2:18 - X10000 as Unlimited Storage Backend
3:05 - Business Benefits and Next Steps

Key Quotes

0:40 "The IT departments managing these environments need the AI dev architecture to be as easy to manage as possible without blowing up the budget."
1:18 "Single store's hybrid transactional and analytical processing database or HTAP allows multiple types of data to be managed under one database umbrella."
2:53 "This helps single store by ensuring that the cluster doesn't need to scale as quickly to make room for more hot data storage just to keep up the query speeds."
3:17 "Customers with sensitive data can keep it in house on the X10,000 where they're in control and still have the option to pay as you go just like the cloud."

FAQ

What is SingleStore's unlimited storage and how does the X10000 support it?

SingleStore's unlimited storage feature is designed to house very large datasets on S3-compatible object storage buckets. The X10000 acts as the on-premises S3-compatible backend, providing all-NVMe performance to reduce latency when queries are directed to that cooler data tier.

Do organizations need to move to the cloud to use this solution?

No. The HPE GreenLake platform provides a pay-as-you-go consumption model similar to the cloud, but the data remains on-premises on the X10000, giving organizations full control over sensitive data while still benefiting from flexible, cloud-like economics.

Categories:

Tags:

Show more Show less

TL;DR

Summary

Chapters

Key Quotes

FAQ

Verge.io: Can You Afford Your Next Storage Refresh?

Bridging the SaaS Protection Gap: Preventing Data Loss and AI Missteps

Unveiling the AI-Driven Underworld of Automation's Rapid Rise

Unseen Data: The Blind Spot in Your Protection Strategies