What you will learn: Initial cloud storage use cases were primarily as backup repositories and long-term archives....
A growing use case is cloud gateways that act as an intermediate step to the cloud. In each of these examples, the cloud is used for the second copy of the data, which means on-premises storage costs aren't reduced. This tip asks you to consider those workloads in your environment that can use the cloud as a primary storage area instead of a secondary one.
The cloud is certainly ready to be used for more than a dumping ground for old data or backup copies of data. Bandwidth, access and security issues have been addressed to the point where interactive access to first-copy data is viable. Doing so requires an understanding of the workloads that should take advantage of cloud storage and how to modify those applications so they can do just that.
A greater return on investment can be derived from cloud storage when it's used as the sole storage area. In this case, data is created directly in the cloud, protected in the cloud and archived in the cloud. Little to no storage is bought for on-site use, which means no resources are consumed for power, cooling and floor space. The applications that could use cloud storage as a primary destination are those where the latency of cloud data access will have limited impact on application performance and user experience.
Use cases for primary storage in the cloud
1. Applications that you want to run on-premises, but that may make more sense to store in the cloud. These are typically applications that create their own data sets driven by sensors placed on devices, security cameras or audio systems. Referred to as the "Internet of Things," most of this data will be processed in the cloud anyway or accessed from endpoint devices where the latency of the cloud connection is less of an issue.
When data will be processed in the cloud, it makes more sense to store it there as it's created instead of moving it later. This is typically because the application receiving sensor or log data will be stored in various cloud data warehouses, like Amazon Redshift, to process and analyze that data.
This use case isn't limited to sensor data. Many organizations find their database applications also work well in a situation where the application is executed on-premises but the database itself is stored in the cloud. The value is that this data is immediately accessible to the cloud compute-powered data warehouse.
2. Data types that are typically accessed by a variety of endpoints for presentation, such as video or audio data. While this data is typically created by a local device or service, putting it immediately into the cloud makes it instantly accessible by these endpoints.
In both of the above cloud storage use cases, the data manager should use object-based protocols via RESTful APIs that can write data directly to the cloud. This can be done with extensions to the database application or with direct extensions to the software code that manages the video and audio data. Typically, this same code is responsible for streaming data back to the above-mentioned endpoint devices, or moving data to the cloud-based data warehouse.
Both of these situations require access to the application code. If that access isn't available, then a gateway may be used. In this case, the gateway’s on-premises storage should be relatively small, primarily acting as a cache to get the data up to the cloud. It may also require a separate application, probably running in the cloud, to deliver this data to the data warehouse or endpoints.
3. File sharing in which the cloud may be preferable to using local, on-premises network-attached storage (NAS). This is because the most common request from users is the need to access file data from anywhere and any device without having to go through a virtual private network connection.
IT, on the other hand, still needs to keep control over of this data. There are now an increasing number of enterprise-class file sync-and-share options -- like those from Barracuda, PanTerra Networks and Soonr -- that store and share user productivity data but still provide exceptional IT control.
Armed with these types of offerings, small to medium-sized IT departments may want to consider the on-premises file server a relic of the past. While NAS systems still have value as on-premises devices, they should be used for much larger data sets or for application and virtual machine data, rather than simple office productivity files.