Panasas’s recent announcement of the addition of an edge platform to its family of scale-out NAS came alongside a planned significant move towards the cloud for the company, with native array S3 support coming later this year.
That adds to moves made in recent months, with the addition of cloud migration, data protection and archiving with Pan Move, and discovery and reporting via Pan View.
The headline announcement earlier this month was the ActiveStor Ultra Edge platform, a high-performance computing (HPC) and artificial intelligence (AI)/machine learning (ML) storage array set of products aimed at edge deployments.
ActiveStor Ultra Edge hardware comes as director and storage nodes, with one of the former to three of the latter in a cluster, which can make for capacity of around 0.5PB (petabytes).
Key workloads targeted are those where customers want to process data nearer to where it is collected at the edge, without having to move it to central datacentres.
It also targets those with smaller capacity needs than serviced by the company’s existing arrays – ActiveStor Flash and ActiveStor Ultra XL – which tend towards much larger requirements, said Jeff Whitaker, vice-president of marketing and product strategy at Panasas.
“We had a number of customers that wanted an entry point smaller than the half a petabyte we had,” said Whitaker. “A lot of customers with remote locations don’t need the capacity but do want the throughput. They were existing customers or new customers with smaller sites or smaller capacity needs.”
Whitaker also said Panasas plans to announce native S3 support in its arrays later this year. This would add to moves towards the cloud announced earlier in 2023 and late 2022.
Here, the company launched Pan Move and Pan View, which are the result of collaboration with French data protection specialist Atempo. Pan Move allows for data staging, copy, move and sync between on-premise and cloud locations. Pan View supports data discovery, profiling and forecasting.
Native S3 support in its array products will allow customers to store data in S3 object format in Panasas hardware and smooth replication to the cloud.
Previously, Panasas had a position of eschewing the cloud, pointing out that while some AI and ML development takes place in the cloud, there are performance issues that many of its customers have to deal with.
“I thought we needed a cloud story; or more than a story,” said Whitaker. “People are doing storage in the cloud – usually in small datasets because scaling I/O in the cloud is costly. It costs something like seven to 10 times [more] to get the I/O performance you need in the cloud.”
“There’s also a network performance bottleneck. We’ve got one customer that has to contact their cloud provider because their compute workload will affect all the others when they run something like 250,000 cores.
“The first step is to allow customers to use the cloud as a cache with persistent data in a datacentre,” he said.
“We will announce native S3 support this year. We’ve been sitting on it, but decided we needed to move because a lot of AI/ML work is being done in the cloud.”
Whitaker said, however, that the company has no current plan to offer its PanFS storage OS-based products to run on the cloud.
“We’ve also thought, ‘Do we run PanFS in the cloud?’, but have decided not now. If there’s customer demand, then we will.”