There are three types of primary cloud storage that I see emerging and all of them use some sort of hybrid approach so that most active data is available locally on an appliance and then sub-sections of that data are only available via the cloud. It is the job of the cloud storage software to be as accurate as possible in making sure that the data you want is locally available. The three use-cases for primary cloud storage are file sharing, collaboration enablement, and semi-structured environments like email.
One of the key reasons that primary cloud storage will still need storage administration is that it will require a local infrastructure that connects servers to that hybrid appliance. Most of these hybrid appliances connect to the local servers via an IP based protocol, typically using CIFS, NFS, or iSCSI. This connection is a storage network and will need to be managed just like any other. The storage administrator will need to make sure that the network can deliver storage to the users and servers as fast as those applications can demand it. In fact to some degree there may be even more pressure on the network even though the bulk of the storage is remote.
As we discussed in our webcast "Adding NAS to Your SAN By Leveraging Cloud" thanks to the intelligence of these hybrid appliances the local store can be relatively small and organizations can cost effectively leverage solid state technology as the local storage area. That means that locally accessed data will actually be delivered faster and as a result the network needs to perform better to be able to match that performance. Local network as well as WAN optimization skills will be an important requirement for the cloud storage administrator.
The other aspect of primary cloud storage is that storage management becomes more important, especially as you're paying by the GB per month for the capacity used. While the expansion of storage is simplified, automatic in most cases, with primary cloud storage you still want to be able to control growth. Though a few providers offer the ability to tier backend storage at the provider’s data center, most do not simply because they have not invested in the integration of an alternative tier like tape. There is typically only one tier and it is often inexpensive disk with a lot of redundancy. As we discussed in our article “LTFS’ Role In The Cloud” cloud providers may want to expand their use of tape beyond backup and start looking at tape as an alternative storage area.
For now, in most cases, you only have one tier, other than the local hybrid tier which is really more of a cache, available to you when using primary cloud storage. With only one tier available to you may want to look for alternative ways to save money as data becomes less important. Moving data management beyond simple tiering requires that you look at data cost reductions in new ways:
-- What data is extraneous and really should be deleted
-- What data is important but does not need to be stored in multiple regions
-- What data can be easily recreated, meaning it does not need to be stored at all
--What data should be shifted to a slower, less expensive cloud storage provider, essentially tiering between providers.
These are all aspects of the new era of storage management. Primary cloud storage brings greater emphasis to managing the cost of capacity. That will have greater importance in the cloud era, and how you achieve maximum cost efficiencies will be a key responsibility of the cloud storage administrator.
Follow Storage Switzerland on Twitter