As we discussed in our last entry, deciding when to use cloud storage is an important decision that data center mangers are now trying to make. While adding the capabilities to your own software was an early approach most were waiting for an easier way to get to the cloud and it comes in the form of software that allows near seamless access. We call these onramps.The onramp software do the same basic core functions, translating the storage protocols that you are accustomed to today and converting them to the storage models that the cloud providers are using. Almost all of the onramps use a hybrid model that leverages a local storage capability and then as bandwidth is available copies that data to the cloud. The onramps typically come in two forms today, either a specific piece of hardware that is provided to you or as a software application that you load onto a server or into a virtual infrastructure.
On premise part of the cloud implementation typically acts as a cache. When data is accessed, written or modified it is stored in the cache for faster access. The data is in most cases copied as soon as possible to the cloud for redundancy. Then like any other type of cache as that data ages and becomes unused it is moved solely to the cloud. This allows for the local cache to be relatively small.
The most common use case of these devices or software extensions is as a part of a backup or archive applications where data is written to a local device quickly so the backup job can complete and then replicate through the cloud as bandwidth allows. This provides an automatic off-site vaulting capability that many users need. Most leverage deduplication and compression to optimize bandwidth, the only downside is the initial seeding of the cloud storage target. Providers and backup software developers have worked around this by shuttling an initial seeding unit to build the baseline by which data optimization can work against.
Onramps are not just limited to backup or archive data. Primary storage onramps to the cloud are now available for a variety of primary storage situations including both NAS and SAN. In cloud enabled NAS product, data is typically moved at the file level in and out of the cache. The SAN type of implementations have to take a finer level of granularity and need to move data at a sub-file level, typically a block. They also tend to have much larger cache capacities than the file serving versions but in general tend to respond to demands very quickly, even within databases. In both NAS and SAN there is not an initial seeding option, so these implementations are ideal for new projects or where data can be slowly populated to the device.
Most users are going to dip their toes in first with cloud storage and onramps allow that to happen with relative ease. For example they can be a device that backups are sent to first, then a storage area for blob oriented databases like Sharepoint or Exchange. From there as confidence builds they can move into the role of providing more traditional primary storage services.
Track us on Twitter: http://twitter.com/storageswiss
Subscribe to our RSS feed.
George Crump is lead analyst of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. Find Storage Switzerland's disclosure statement here.