Wednesday, January 19, 2011

Cloud Architectures For Success

The key driver for cloud storage solutions that are ready for deployment in 2011 will be those that use a hybrid, on-premise / cloud deployment model. This allows IT managers to leverage the abilities of cloud storage to finally create a tiered storage strategy without the side effects that have prevented adoption in the past. An important ingredient for success is that the users see a single view of all the data. Users historically have not had the time or discipline to manually copy data from one storage area to another. This movement between the local data set and a cloud storage location needs to be automatic, essentially cache-like. A hybrid architecture creates an on-premise appliance that caches data locally and then, over time, replicates that data to a cloud storage provider. Because it is a cache the volume of local storage remains relatively small, and growth happens in the cloud. This eliminates much of the time spent managing local storage because tasks like capacity expansion and system upgrades are now the cloud provider’s responsibility.

As stated earlier this process also eliminates the need for backup of any cloud based data set. In this architecture data is replicated to the cloud almost as soon as it is created or changed, and as a result it’s protected sooner and probably more securely than by an internal backup process. Also, because the appliance can be easily recreated and started, with the correct authorization, disaster recovery can happen from almost anywhere.

Additionally, because the cache is relatively small and it doesn’t grow (or grows slowly), the investment for it can be made on higher performing storage, like solid state. This allows for a very cost effective use of a high performance but expensive technology, causing users to actually see a performance increase because of cloud storage.

For the hybrid architecture to make sense though, the use case for a cloud storage system has to be right. The ideal use case will be a data set that can be easily segmented between active components and inactive components. The active components are stored on the cache with a backup copy sent to the cloud storage provider. The inactive components would be stored only in the cloud. The most obvious example of this use case is a file server.

A cloud enabled NAS like the Nasuni Filer addresses one of the biggest challenges facing data centers of all sizes; the rate of storage growth, both in terms of capacity and in number of servers. Compounding this is a perceived need to retain discrete file data like those created by office productivity applications. These files are increasing in size, increasing in number and increasing in importance to organizations. The result is an environment that is near fatal to backup applications, millions and millions of files where 80% or more never change but must be retained to satisfy user “just in case” demands or to address specific industry retention requirements.

Without a cloud enabled solution customers are more frequently leveraging another popular technology, server virtualization, to address their needs for file services. In the virtual server environment it appears easy to start up file server after file server to satisfy user demand. The problem, in addition to server sprawl, is that using an entire OS load to manage a very specific task like file services wastes resources which are particularly valuable in the virtual environment. An alternative for many is to select a NAS, but as we indicated in our article “How To Add NAS To a SAN” this is typically overkill for office productivity files.

Using the hybrid cloud architecture to provide these types of services is an ideal next step for cloud storage in 2011. To leave a “known” like traditional file servers, and try something new requires a significant potential upside versus the status quo. Using the cloud for file services provides that upside by removing one of the biggest challenges facing the data center, managing the growth of file data while improving protection, disaster recoverability and long term retention. The virtual appliance approach allows for easy download and setup and provides complete file services via that appliance. Leveraging both the existing virtual server infrastructure and SAN infrastructure, the Nasuni Filer keeps cost down and allows an IT administrator to move quickly from install testing to full production by performing a simple VM migration.

George Crump, Senior Analyst