tashatuvango - Fotolia
Moving data to cloud storage isn't a trivial task, but by now, most IT administrators know how to do it. But moving data from cloud storage to another cloud or to an on-premises system is more difficult and requires careful cloud migration planning.
There are plenty of free and low-cost services for consumers and individual users that enable fairly simple movement of files to the cloud, from the cloud to on-premises infrastructure or from one cloud to another. These services work well for users with limited amounts of data. But moving data from cloud storage when there are many users, structured or large amounts of data, limited time windows and intact security, the processes are neither simple nor inexpensive.
Moving sticky data
Once data moves into cloud storage, it has a tendency to stick. There are impediments to moving that data out of cloud storage, such as transit fees, which are charges for data read and moved out of cloud storage. Also, large amounts of data generally require cloud migration planning. And incompatibilities between different public cloud storage vendors and on-site private clouds can cause issues.
For example, the API for Amazon Web Service's (AWS) Simple Storage Service is the de facto standard for both public and private cloud storage, but not all S3-compatible APIs are identical. Most compatible S3 APIs are subsets of the original AWS S3 API. Not all applications or data migration software will work with all versions of the S3 API.
So, what is the best way to move data out of cloud storage? It depends on how the data moved into the cloud in the first place. If the data was created on a virtual machine instance in the cloud or migrated to the cloud, it might be simplest and fastest to replicate that VM to an on-site or other cloud VM. Doing so requires that the destination hypervisor and VM be the same as the source or that there is software that makes the conversion in flight to the destination hypervisor and VM type.
Cloud migration planning pointers
Hypervisors typically have VM replication options built in, but they work only with variations of their own hypervisor. VMware vSphere has the most options, including VM Replication, Storage Vmotion, Site Recovery Manager and compatibility with a slew of third-party vendors. Some of those third-party software options do conversions to and from other hypervisors, such as Microsoft Hyper-V to Azure.
The problem with all forms of VM replication is that they are on a VM-by-VM basis. If there are only a few dozen VMs, that's no big deal. When there are thousands or tens of thousands, it can be a daunting task.
If the data was migrated to the cloud storage via a caching cloud integrated storage (CIS) system or storage system that tiers to the cloud, it would be best to use that system to move the data back on premises or to another compatible cloud storage. Because the CIS or tiered storage system is already linked to the data, it is relatively simple to pull data back or move it somewhere else.
Another way to pull the data out of cloud storage is similar to how many put it in: installing a storage system or device on the cloud storage service provider's data center. Migrate the data to that storage system, and then, delete the data from the cloud storage service provider. Pack up the storage system, ship it to where the data will reside, install the system at its new home, set up that storage system as the new target home of the data for those who need access or migrate the data to its destination storage system. But this is not an easy project; it takes more than just good cloud migration planning to use this method.
In the end, there is no best way to move data out of cloud storage. The right method depends on how the data was put in the cloud storage; what systems, software and services are available to move it out; and what systems, software and services the organization has the most familiarity with.