Geoscience teams in the Oil & Gas industry are increasingly migrating critical subsurface workflows to the cloud to leverage elastic compute, AI resources, and centralized management. However, running demanding geoscience interpretation software, such as S&P Global's Kingdom, in the cloud often forces organizations to make an unacceptable choice between workflow performance and team collaboration.
The Challenge: The Dilemma of Cloud Data Silos
Seismic interpretation involves massive datasets (ranging from hundreds of gigabytes to petabytes) and generates incredibly high volumes of small, serialized I/O. When deployed on traditional native cloud file services, this specific I/O pattern can overwhelm the storage, turning tasks that should take minutes into hours and completely stalling a geophysicist's ability to iterate on prospect evaluations.
To bypass these cloud storage bottlenecks, many teams resort to downloading their seismic data onto local NVMe workstation drives. While this local approach speeds up the processing, it introduces severe business risks: it creates data silos, fragments the "single source of truth," and strips away essential enterprise data protection, sharing, and backup capabilities. It also stalls processing while the data is being copied.
The Solution: A Certified Reference Architecture on Azure
To provide both the speed of local drives and the collaborative safety of the cloud, Qumulo has collaborated with S&P Global and Microsoft to validate a certified reference architecture for Kingdom running on Cloud Native Qumulo (CNQ) in Azure.
This solution pairs Azure GPU VDI workstations (NV-series A10) with CNQ's scale-out shared file system over SMB3. By featuring an all-NVMe performance tier backed by the cost-effective capacity of Azure Blob Storage, CNQ decouples scale from performance, serving Kingdom's latency-sensitive small I/O while safely centralizing data in the cloud.
The Proof: Head-to-Head Benchmarks
In head-to-head technical validation testing using identical VDI configurations and datasets, Cloud Native Qumulo was tested against the native Azure Files Premium v2 SSD service. CNQ delivered dramatic acceleration across three real-world Kingdom workloads:
Process Multiple Traces: Completed in just 1 hour and 14 minutes on CNQ, compared to 8 hours and 2 minutes on Azure Files (~6.5x faster).
GenerateBrickVolume: Completed in 11 minutes and 9 seconds on CNQ, compared to 1 hour and 3 minutes on Azure Files (~5.7x faster).
SMT_LoadTest: Completed in 1 hour and 28 minutes on CNQ, compared to 5 hours and 59 minutes on Azure Files (~4x faster).
During the testing, Qumulo's cluster analytics revealed that CNQ seamlessly serviced every operation with sub-millisecond latency, processing the demanding I/O profile with significant performance headroom to spare.
The Business Impact
For Exploration & Production (E&P) organizations, architecture directly dictates productivity. With Cloud Native Qumulo, geoscience jobs that previously took up to 8 hours now finish in roughly an hour.
This massive reduction in time-to-results keeps interpretation highly interactive and ensures that organizations maximize the ROI on their expensive GPU cloud workstations and Kingdom software licenses. Most importantly, it allows IT leaders to finally eliminate risky local storage silos, empowering the entire geosciences team to collaborate securely on a single, protected, and fully managed source of truth in the cloud.