Luiz - Fotolia
- George Crump, Storage Switzerland
Every data center's storage architecture should include cloud storage. How that architecture integrates or uses it will vary from organization to organization. Strategies to maximize cloud storage benefits shouldn't focus on just one aspect or capability, but instead should consider the whole picture and what cloud storage would actually mean to the enterprise.
Typical initial cloud projects, such as backup, migration, archive and disaster recovery (DR), tend to limit long-term cloud storage benefits and use of cloud resources. For example, a cloud backup application that stores data in the cloud using a proprietary format means an enterprise can't subsequently use that data with another cloud-based application.
Migration approaches, on the other hand, typically transfer and transform data, making it cloud-native and useable by cloud applications in general. Most migration services perform one-way only, however. They don't enable the seamless movement of data back and forth between cloud and on premises and across clouds.
The goals of cloud storage
Organizations should establish their goals for cloud storage before they implement any applications or workloads. They should also use the fewest number of cloud services possible to achieve those cloud storage benefits, as well as ensuring that how data is stored doesn't preclude its access and use by other use cases.
Cloud providers sell two primary components: storage and compute. If an application doesn't store data natively, an enterprise will only use the cloud as a digital dumping ground. It won't benefit from the cloud's other component, compute. IT planners should also aim to move data as seamlessly and as natively as possible between on premises and multiple cloud providers.
IT planners must select a starting point for their cloud journey while also establishing a foundation that supports all desired cloud storage benefits. In addition to mapping out the trip to their destination, admins must select the vehicle that will take the organization to that destination. There are important landmarks to consider along the way.
The importance of cloud-native data
Making data cloud-native is critical to moving beyond just using the cloud as a digital dumping ground. Storing data in a cloud-native format enables the use of cloud resources for DR, analytics and even permanently migrating an application to the cloud.
Integrating the cloud into the data protection process provides an excellent starting point. There are many data protection products, both backup and replication, that support copying data-to-cloud storage. However, these products may keep copied data in a proprietary format once in the cloud, making it inaccessible to other cloud applications. If an organization's backup application doesn't store data in a cloud-native, accessible format, it should consider an alternative. Otherwise admins may have to "restore" to the cloud for data to become accessible in cloud-native form to other applications.
Replication tends to copy data in its local-native format. Once in the cloud, the data must be transformed prior to use. Most cloud providers furnish data transformation utilities as part of their service, but the utilities require additional steps prior to making the data readable by cloud applications. An increasing number of replication products, however, transform data simultaneously with the replication's transfer. These also perform data transformations while copying the data between clouds and back on premises.
An alternate approach eliminates the need to transform data by making the file system used on premises, in the cloud and across clouds identical. Companies, such as Elastifile, SoftNAS and SwiftStack, provide cloud file systems or cloud fabrics that deliver a single file system that spans both on-premises and multiple cloud storage locations to maximize cloud storage benefits. The file system manages the movement of data among locations based on user policy or preference.
These file systems provide SMB, NFS and sometimes iSCSI support and enable organizations to use the cloud with legacy applications. IT directs backups to the file system as if it were an on-premises NAS. The file system replicates or moves that backup to the cloud based on policy.
The cloud file system requires greater commitment because IT must move data into it. But once complete it offers more seamless use of cloud resources. Replication with transformation requires less commitment and is merely an extension to existing processes. Both approaches enable a long-term cloud strategy, where data moves seamlessly between on-premises and multiple clouds.
Because cloud storage is only one part of available cloud resources, an organization's cloud strategy should use both cloud storage and compute. As we've seen, many "starter" cloud use cases eliminate cloud compute by storing data in a proprietary format. Replication with transformation or a cloud file fabric makes data accessible to all cloud resources. Organizations can then use those resources to enable DR and cloud-based data analytics and shift applications to the cloud, thereby increasing the benefits they get from cloud storage.