An ROI Every Month


Cloud storage as either a backup target or as an archive target allows customers to off-site data and dissolve on-site storage costs. The downside is that while on-site storage costs are dissolved an ongoing storage cost is incurred as the services are used. As a result any optimization of the data going to cloud storage shows an immediate cost savings each month. Also spending extra effort and processing time to achieve maximum data reduction is well worth the minor performance cost since this data is not typically performance sensitive. In fact the net may be a performance improvement since less data needs to be transferred across the WAN connection.


Cloud storage services can be deployed purely as an outsourced model with no on-premise equipment but what is becoming more common is to use a mixed model. In this model companies like Bycast, Nirvanix and Iron Mountain place an appliance at your site. This appliance looks to the users like a NAS with bottomless storage that supports NFS and CIFS protocols. Moving data to the cloud then is no different then moving data to a local NAS. The migration to the cloud happens transparently in the background. The appliance translates between the user friendly NFS/CIFS and more common internet protocols like webdev or http. The important reality here however is the storage on the NAS, while it appears bottomless it is also billed to the IT department on a monthly usage cost. Deduplication of that storage could reduce the monthly expense by 4X to 10X depending on your data type.


Cloud Storage Needs Agnostic Deduplication


The manner in which deduplication is implemented is critical when trying to optimize cloud based storage. Deduplicating cloud storage can not be done by an all-in-one storage platform. If deduplication is built into the OS or the storage system it will not know how to interact with the cloud based storage. Cloud storage is different and while the interface to it may be via a cloud gateway, the underlying OS is certainly different.


Deduplicating cloud storage then is going to require a stand alone deduplication software application from companies like Ocarina Networks that are independent of the actual storage platform and do not require a specific storage platform to work on, in other words it has to be storage agnostic More importantly this optimization has to have the ability to optimize from any type of storage to any type of storage including cloud storage. Companies like Ocarina take this a step further by not only deduplicating data prior to it being written to the cloud storage, it also can assist with the migration possibly eliminating the expense of an archive application.


Optimizing WAN Bandwidth


WAN bandwidth is another concern in an environment that is using or considering using cloud storage. The amount of data that needs to travel up and down the WAN network needs to be carefully paid attention to. An increase in the bandwidth requirement could lead to a costly increase in WAN bandwidth costs and an increase could be very likely as access to the cloud storage service is broadened across the user community and its usefulness is understood. An upgrade to WAN bandwidth requirements could likely wipe out any costs savings that were derived from the cloud storage.


Extending Cloud Archive to Cloud Backup


Deduplication that can optimize data before going into the cloud will keep bandwidth costs under control. It can possibly also allow these archive only systems to be used for the backup process. When a backup software application creates a backup, especially a full backup, it creates a big file or series of files which contain all the individual files that need to be backed up. This file is then written to disk.


The problem is that while this disk is very similar to the file that contained the previous full it is stored as a net new file. To get around this proprietary deduplication systems were developed to eliminate the redundant data; while these work fine, they can't typically work with another vendor’s cloud storage service. Storage agnostic solution however can.


Cloud Deduplication Architecture


Architecturally a storage agnostic optimizer could either reside on a dedicated appliance that would then analyze file systems that you designate for files that meet a particular criteria, lack of access, size etc.. and then migrate to the cloud storage appliance described above. During this migration the file will be analyzed for dedupability and compressibility for maximum data reduction. At that point only the "thin" file is stored on the Cloud NAS for maximum cost savings.


Optionally providers of Cloud NAS could work with companies like Ocarina Networks to integrate the optimizer into their on-premise component. This would give them a tremendous competitive price advantage in the cost per GB battle.


With either deployment model cloud storage deduplication allows the customer to win and cloud storage is further considered to be a viable and cost effective mainstream alternative to traditional storage.