ra2 studio - Fotolia
SwiftStack's new software upgrade enables two-way data synchronization between its on-premises private cloud storage and the public cloud storage offerings from Google and Amazon.
The SwiftStack object storage software 5 release, which became generally available today, enhances the Cloud Sync feature the San Francisco software vendor added late last year. Cloud Sync enables customers to set policies to automatically place data across private and public storage for off-site data protection and cloud-based archiving. SwiftStack stores data in the same object format both on premises and in the cloud.
SwiftStack object storage is based on open source OpenStack Swift software and runs on commodity Linux servers. The company sells and supports a commercial version of Swift, and its engineering team adds management services and features such as load balancing, metadata search and Cloud Sync. SwiftStack object storage also includes a file system gateway and the vendor is working on native file access capability.
Updated Cloud Sync
The SwiftStack Cloud Sync feature formerly supported only one-way replication to a bucket in Amazon Simple Storage Service (S3) or Glacier or in the four Google Cloud Storage offerings. The new bidirectional replication capability is designed to ease collaboration with external partners by synchronizing on-premises data with shared cloud buckets, and to facilitate bursting to the cloud for additional resources.
"Before, if people were going to do a lot of compute, they had to more manually replicate data from a SwiftStack cluster to a spot in the public cloud and then replicate the results back," said Mario Blandini, vice president of marketing at SwiftStack. "Now that we have this going both ways, it allows you to use that elastic compute in the public cloud."
Scott Sinclair, a senior analyst at Enterprise Strategy Group, said the new Cloud Sync capabilities would also improve collaboration by synchronizing data that users and applications modify within the public cloud back to the on-premises storage. He said the new capabilities would help SwiftStack to "transition from an on-premises software-defined infrastructure," with tiering to the cloud, "to more of a hybrid cloud storage layer.
Scott Sinclairsenior analyst, Enterprise Strategy Group
"With this next step and the bidirectional aspects, the cloud infrastructure becomes a partner in the data center ecosystem," Sinclair said. "And if you follow this trend moving forward, there's the potential for [SwiftStack] to start delivering a more capable and more functional hybrid cloud infrastructure."
Steven Hill, a senior storage analyst at 451 Research, said the value of Cloud Sync replication was illustrated by the three-hour outage that Amazon S3 storage services experienced on Feb. 28 "and the resulting chaos for websites and customers who depended solely on the S3 East Zone out of Virginia.
Hill said a website using SwiftStack object storage with Amazon S3 and Google Cloud Storage could have simply pointed to on-premises storage or the Google Cloud Storage when Amazon S3 became unavailable. That would, however, require customers to pay to store the same data in two clouds.
"The storage would have existed at more than one location. The challenge is keeping all those different resources in synchronization," said Hill, noting that that's what SwiftStack's Cloud Sync does.
Upcoming SwiftStack features
SwiftStack object storage does not support direct bidirectional replication between two public clouds. Blandini said a future SwiftStack release would support synchronization from one public cloud to another public cloud for new "multicloud" use cases.
"Our vision has always been that we would [provide] data management for data that resides on prem or in 'n' number of public clouds," Blandini said.
Another upcoming feature Blandini mentioned is "policy balancing" to enable the system to put more data in one cloud versus another cloud if, for example, the cloud storage service prices were to change. The customer might have one or more copies of the data on premises and another copy in Google or Amazon, or distributed between the two public cloud services. SwiftStack plans to add policy balancing in the next four to six weeks, according to Blandini.
In the meantime, other new capabilities in SwiftStack 5 include multiregion erasure coding for data protection and support for deep buckets with more than 100 million objects from the OpenStack software distribution. SwiftStack is also making available a new desktop client for the Windows and Apple OS X operating systems.
Laura DuBois, group vice president at IDC's enterprise storage, server and system infrastructure software research, claims multiregion erasure coding is a "must-have" for any object-based storage.
DuBois said SwiftStack's main challenge is "go to market" against the major vendors of object-based storage such as Dell EMC, Hitachi Data Systems, IBM and NetApp.
"What sets [SwiftStack] apart is the file/object gateway, flexible deployment and OpenStack integration," DuBois wrote in an email. She estimated that approximately 10% of SwiftStack's customers use OpenStack, and many hail from the media and entertainment and life sciences industries.
Swift or Ceph: Which is the best choice?
OpenStack Newton release includes encryption
SwiftStack and Cisco team up for object storage