Hammerspace adds S3 interface to Global Data Platform
Hammerspace adds the S3 interface to its parallel global file system, allowing object storage to talk directly to the software while bringing more efficient data use to compute.
Hammerspace is adding object data to its Global Data Platform with the addition of the S3 interface, enabling customers to unify siloed data regardless of where it is or what format it's stored in.
The addition of the S3 interface brings support to all protocols, adding object storage to the already supported file and block formats. This helps customers with orchestrating data sets to available compute resources, which the vendor highlighted as being beneficial for high-performance computing use cases and GPU processing.
Prior to the addition of the S3 interface, Hammerspace used S3 as a storage target, but cloud-based applications that use S3 couldn't write natively to its Global Data Platform, a parallel global file system. With the addition of the S3 interface, Hammerspace customers will be able to move files from the Global Data Platform into object storage and back.
The ability to move data from file to object storage is what customers want, so they can bring the data to the compute for analysis, according to Mike Matchett, an analyst at Small World Big Data.
Now you can be fairly fluid about giving a compute cluster access to your data across your organization, no matter where it is.
Mike MatchettAnalyst, Small World Big Data
"Now you can be fairly fluid about giving a compute cluster access to your data across your organization, no matter where it is," he said.
The Global Data Platform can provide a view and add metadata tagging across NAS systems, object stores and cloud-based S3 repositories to allow for fluidity from one to the next, Matchett said.
S3 for GPU analysis
Hammerspace has seen an uptick in adoption of its data orchestration platform in recent years to get data to compute nodes, according to Molly Presley, senior vice president of marketing at Hammerspace. Now, Hammerspace is making data created by S3-based applications accessible to customers.
Hammerspace's orchestration technology works by unifying and assimilating metadata. As users look to process data, the metadata is moved to the compute in real time; when it is time to analyze data, Hammerspace prioritizes the order, moving data that needs to be processed first instead of moving all the data to the local compute, Presley said.
But moving large data sets around is no trivial task, according to Ray Lucchesi, president and founder of Silverton Consulting. Hammerspace is moving the data without using resources for data analysis and without customers having to find ways to bring the data together. Regardless, bringing data together that's stored in different geographic regions can be a slow process.
"It's a nice capability, especially when you're moving lots of data," he said, "but you're going to pay the performance penalty, regardless of where it is."
Adding to, not changing
Adding S3 integration is an extension of how Hammerspace handles data, according to the vendor. In the past, objects could be inserted into the Global Data Platform, but Hammerspace didn't have the ability to granularly see into the data or utilize the metadata the way it could with NFS and SMB.
Now, Hammerspace can write an object from S3 the same as it would a file, Presley said.
The addition of object storage adds another level of flexibility to the storage and data landscape, Matchett said. That is important if a customer has different centers of computing and needs to orchestrate movement to those centers.
"Adding object storage protocols to their global file system solution is a great advancement and opens up lots of use cases," he said.
Adam Armstrong is a TechTarget Editorial news writer covering file and block storage hardware and private clouds. He previously worked at StorageReview.com.
Dig Deeper on Storage system and application software
