This is for AI training cluster. The customer is using DDN as a default de facto standard for their AI cluster.
Enhancing AI cluster training with reliable performance and good customer service
What is our primary use case?
How has it helped my organization?
DDN is one of the storages actually supporting by NVIDIA Superport. The customer is using NVIDIA Superport, which helps in terms of scalability.
What is most valuable?
The DDN Exascaler uses a parallel file system, Lustre, which is proven for HPC and AI workloads. The performance is outstanding, and customers are very happy with it. It is a reliable system.
What needs improvement?
There are challenges with the operation cost and upgrading process, which is disruptive, especially with the node. Some customers have moved away due to these operational issues.
What do I think about the stability of the solution?
Customers are happy, especially with the performance numbers. They are using DDN as a reliable system for their AI clusters.
What do I think about the scalability of the solution?
They can currently scale up to 50 petabytes, which is massive.
How are customer service and support?
I'm not seeing real cases of complaints from the customer side. People are happy with the service.
How would you rate customer service and support?
Positive
Which solution did I use previously and why did I switch?
We are using DDN Exascaler. We are not using DDN Fusion.
What's my experience with pricing, setup cost, and licensing?
The cost is higher compared to other offerings, but the performance is good. The challenge in comparison to other offerings is that DDN is a little bit expensive.
Which other solutions did I evaluate?
Their main competitors include Weka, NetApp, IBM, and WAST. Some customers have moved from Lustre to other systems due to operational issues during the upgrade process.
What other advice do I have?
DDN should not make my information public. Mention me as a chief architect without naming me or my company.
I'd rate the solution eight out of ten.
