Artificial Intelligence (AI) will have the largest impact within IT infrastructure more than any other technology over the last few decades and has transformed our lives in ways unmatched by other generational previous technology waves. However, the broad use of AI requires a new type of data architecture that’s different than today’s storage and server products. To generate effective AI results, it requires far more processing power than previous generations of IT systems, and performant storage is needed to feed these higher performing servers. 

Along these lines, StorageReview has completed multiple rounds of testing on Myriad.  The first review was in November 2023 and covered the foundational elements. Since then, Myriad has been updated with new features and performance tuning, optimized for a new generation of workloads including AI applications. This review showcases how Myriad meets the demands of new workloads with the performance, scale, and ease of use needed for the upcoming wave of enterprise demands.

“Quantum Myriad storage cluster exhibits exceptional performance in managing diverse and demanding I/O operations. It is a versatile solution for traditional business workloads and cutting-edge AI applications.”

New Workloads Including AI Applications

AI workloads are different than traditional applications. These new workloads are far more graphics intensive, as many AI models utilize the latest GPU (graphical procession units) from NVIDIA coupled on high-performance servers to process the data yielding new AI language and inferencing models.

This is precisely the type of testing performed by StorageReview to better assess Myriad’s capabilities for these new workloads. StorageReview’s test system was configured with state-of-the-art GPUs and high-speed systems, typical of real-world systems. 

Besides overall performance, the new generation of workloads require the ability to scale easily, as enterprise demands are no longer static; they ebb and flow to meet the ever-changing needs of the business. In addition, enterprises are short in staffing, and next-generation platforms must be highly automated for easier operation and scale. StorageReview was able to test these characteristics as well. 

Results and Key Takeaways

The lab testing was highly successful, and the full review gives many highlights. We highly recommend reading the full report to better understand Myriad’s capabilities for this new generation of workloads. There were a few positive surprises from the test team, including how well Myriad scales with performance across multiple nodes simultaneously. 

Overall, the StorageReview team was impressed with the performance and commented “This indicates excellent performance for large-scale data processing tasks common in machine learning model training, where large data sets are frequently accessed.”