Scality Ring 9 brings Storage Accelerator enhanced flash tiers

0
212
Oracle enhances customer experience platform with a B2B refresh

Source is ComputerWeekly.com

Scality has announced version 9 of its Ring object storage software. Key additions include enhanced tiering – called Storage Accelerator – for high-performance use cases, new monitoring tools based on the Prometheus open source product, and new integrations with VMware and backup software supplier Veeam.

Flash support was added in Ring 8, with some tiering possible. In Ring 9 Storage Accelerator, tiering is built on a combination of media type and data protection method.

So, for example, data with very rapid access time requirements could initially be held on NVMe flash and three copies made. Without erasure coding, that would mean a 200% capacity overhead, which may be deemed acceptable for the most mission-critical data.

As that data aged, it could be moved according to a time trigger to a lower tier of flash, such as QLC, and making use of erasure coding, which would give an overhead of just 33%  as data is shared across locations.

Finally, data could be tiered off to cloud or tape for longer-term retention.

“What we’ve done is to allow tier one workloads that make it fast enough to consolidate from NAS onto Ring,” said Paul Speciale, chief marketing officer at Scality. “We have hospital customers who use Ring for medical imaging, but could move such workloads with rapid access requirements to it.”

Meanwhile, Scality’s new Ring monitoring tools are built around the open source Prometheus platform. Prometheus uses Restful APIs and supersedes legacy SNMP-based tools which were used previously in Ring as a management UI.

Customers can use Prometheus Rest APIs to query the health and performance of system components. It also has database functionality that allows logged data to be analysed via time series data using a query language.

Prometheus APIs can be used to query Ring endpoints and also come with an alert manager.

Will Scality use its monitoring tools as part of a consumption model of storage deployment? Some suppliers have incorporated storage estate monitoring with built-in triggers that give the option to deliver more storage or performance according to monitored bottlenecks or resources nearing capacity.

Speciale said there were no immediate plans to do so. “We have our own remote monitoring tool used for support,” he said. “And we collaborate with HPE Greenlake, so any work on consumption models is based around Greenlake. AIOPs is an area we will be looking into.”

Elsewhere, Scality has added a VMware object storage extension. This uses calls to manage, monitor and control Ring object storage from VMware vCloud Director. So, for example, customers can use it to provision object storage in Ring, to create a bucket, to create policies on a bucket, and so on.

Scality object storage APIs for monitoring, reporting, data placement, and so on would be added for Veeam, but not until 2023, said Speciale.

Ring 9 also adds full Amazon access controls.

Scality’s Ring object storage runs on relatively inexpensive commodity server hardware. Ring also supports NFS and SMB and provides file access storage.

“People buy Scality Ring and grow with it, such as starting in backup and archiving and growing from there,” said Speciale.

“Usually the buyer is enterprise IT and they’re running a range of systems such as SAN and NAS and are often VMware-based, although there is a noticeable turn to Kubernetes with Red Hat OpenShift deployments for Splunk [in Kubernetes].”

Speciale said Scality has banking customers running Splunk for fraud detection with 100s of PB analysed for anomalies and 100TB of new data added daily.

He also cited Germany-based travel services provider Amadeus, which runs Splunk analytics on 14PB covering two weeks’ worth of data, with that updated daily with a further 1PB.

“They’re using it to work out popular destinations, how to price things, etc,” said Speciale. “Ring is like an internal cloud – like an on-prem version of Amazon S3.”

Source is ComputerWeekly.com

Vorig artikelExpensive datacentre outages: Untangling messy collaborations, contributing costs and complexity
Volgend artikelG-Cloud 13 launch hits shaky start as suppliers complain about missing framework functionality