12_tribes - Fotolia
A closer look at cloud storage use cases, strategies
Find out how primary storage vendors are closely integrating with cloud storage providers in ways that enable new uses for cloud storage.
Cloud compute and cloud storage have evolved from overhyped concepts to legitimate options for IT professionals addressing real business challenges. In addition, traditional on-premises primary storage vendors are delivering practical integration with cloud storage providers.
Architecturally, primary storage providers use several approaches to extend their storage systems' reach into the cloud:
- Place matching systems on-premises and in the cloud provider's data center.
- Enable the primary storage system to send data to a dissimilar secondary storage system, including cloud storage.
- Create a virtual instance of the storage system in the cloud provider's environment. This uses the raw storage capabilities of the cloud provider, but management is performed with the storage vendor's software.
Each of these architecture decisions can enable various cloud storage use cases, which are explained below.
Use case 1: Cloud as DR
The most common use case for cloud storage is as a disaster recovery target. Many vendors now provide the ability to replicate data to a cloud provider to protect against a data center outage. In most cases, this capability uses a virtualized instance of the vendor's storage software running in the cloud. Replication software essentially "sees" another storage system in the cloud and replicates data to it. The other option is to set up an identical storage array in a cloud data center. Unlike the virtual instance, this hardware can be dedicated to the subscribing organization.
It is important to note that both of these use cases simply provide data movement from point A to point B. It does not mean that the applications and networking needed to access those applications are properly architected. In theory, both options could provide that complete functionality, but that is a process the organization needs to work through with the provider.
If an organization wants to completely fail over to the cloud, it will need to work with the cloud provider to ensure that appropriate network conversions are handled automatically in the transition. Additionally, it must be sure the provider has the available compute resources and the ability to start that application in their cloud when needed. Finally, the organization should ensure the provider can deliver the appropriate quality of service while the application is running in its cloud.
Protecting the cloud
As more data is created in the cloud, protecting that data becomes critical. One way to do this is to copy data to another cloud provider. Or secondary copies could reside on-premises. For example, it might make sense to copy archive data to an on-premises tape device as well as the cloud. If you are developing an application in the cloud, it may make sense to use cloud-to-cloud backup software. There are a lot of options available to ensure data that lives in the cloud is protected. Don't make the mistake of assuming your data is safe because it's in the cloud.
Use case 2: Cloud as a tier
Many storage systems today have the ability to move data from one tier of storage to another based on user-defined criteria. This is often the movement of data between high-performance flash storage to slower, less-expensive hard disk drive storage.
By integrating a cloud storage connection, the tiering software can also move data to the cloud. The movement between disk and cloud is seamless, other than the latency of the Internet connection. This approach may be preferable for archiving data, since users can set policies around data age, usage and so on.
In most cases, the performance difference between cloud and disk is negligible for a couple of reasons. First, many organizations have invested in high-speed Internet bandwidth and, second, access would be limited to occasional retrieval of discrete files. As a result, the time lag in transfer would hardly be noticeable in most situations.
Use case 3: Cloud as an application incubator
One of the most expensive aspects of developing a new application is the cost of the hardware associated with that development. Typically, the storage system selected for development is actually the system that will be used during production, so it must be fully scaled to the performance and capacity demands of that application in its production state.
The cloud may be a better place to start application development. Compute and storage can be purchased as the process develops instead of all upfront. Also, the application can be better stress-tested by temporarily leveraging the massive compute and storage available in the cloud prior to bringing the application into production. Finally, developing in the cloud allows the purchase of the on-premises equipment to be far more accurate. The key ingredient is how to move this application from the cloud to on-premises storage for actual production use.
Some primary storage vendors accomplish this with virtualized cloud versions of their storage systems. This allows the application to be developed and tested on the exact same storage software that will be used in production. Using a virtualized version of the storage system makes transfer to the on-premises physical system far easier. The application and its data can be replicated or migrated between the two systems.
Most organizations will never reach a point where they are comfortable running critical applications in the cloud. This hybrid approach allows organizations to use cloud when it makes sense, while continuing to maintain a data center for predictable performance and high security.
Use case 4: Cloud as primary storage
Storage system vendors now have the ability to host unstructured data in the cloud. This hybrid approach uses an on-premises appliance that caches data locally, while the primary copy of data is located in the cloud.
An increasing number of vendors offer the ability to host block storage in the cloud. These appliances use sophisticated capabilities to make sure that frequently accessed data is kept on-premises. The result is a relatively small but active segment of the data set stored entirely on flash storage. Applications like SharePoint and Exchange are particularly well-suited for this. The cores of these semi-structured environments are relatively small, but attachments increase the size of the store. Also, both applications have APIs that allow vendors to safely move older attachments to the cloud.
Today, most storage systems and IT professionals treat the on-premises data center as the hub of the data universe with the cloud being a spoke. In the future, this model may be turned around so that the cloud is the hub. This will be especially true for organizations with a highly dispersed workforce.
George Crump is president of Storage Switzerland, an IT analyst firm focused on storage and virtualization.