Cloud storage brings development opportunities for clustered NAS markets

Cloud storage is now no new concept. And, in fact, for users, there is no need to understand the details of this concept. From a consumer's point of view, it is actually a service that provides a storage business. This, however, puts a higher demand on cloud storage service providers to some extent. On the one hand, cloud storage service providers need to provide perfect support for each user, ensure access performance for hundreds of or even thousands of users, service-level agreements (SLAs), and, on the other hand, cloud storage service providers themselves to focus on storage costs and ease of management, That's one way to cut revenue in some ways.

Cloud storage business model

If cloud storage is a convenient and low-cost storage option for users, these storage challenges can be said to be passed on to the cloud storage service provider.

In this case, almost all cloud storage vendors are turning their eyes to cluster NAS.

What is cluster NAS?

In general, cluster Nas is a loosely coupled collection of compute nodes. Loosely coupled here, in fact, the basic concept and the concept of software engineering is somewhat similar. In cluster NAS, the smallest unit individual is a file, not a data block, and each node is a separate entity with one or more volumes that cannot be accessed by other nodes. When a user stores data, the data is stored in a file form on a node. For data protection, there may be copies of files on other nodes of the same cluster, but they cannot be made available to users in the form of file instances. In other words, cluster NAS has no special requirements for the interdependence and order of nodes. This is different from the cluster system which is operated by the data block mode.

This kind of working method has high practical value to the large application, especially the continuous file data stream. Through file system scheduling algorithm, it can isolate the whole application into smaller and parallel independent tasks, and allocate file data to each cluster node. At present, it is commonly used in large database query search, intensive computing, life science, energy exploration and animation and other fields.

For cluster NAS, hardware is just one part, and file systems and applications are also valuable components. The cluster NAS offerings we can often see include BlueArc's Titan and Mercury series, HP's Ibrix series, IBM's Sonas, Isilon IQ series, NetApp Ontap GX, and so on.

How does a clustered NAS architecture adapt to cloud storage requirements?

This year's cluster NAS market swept past the doldrums, looking like a booming scene, mainly thanks to some of the big manufacturers, such as Dell, such as HP, such as IBM.

So why are these manufacturers willing to relate to cluster NAS? We can get some answers from IBM's vision of Sonas. In the communications with IBM systems and technology's Greater China store manager, he first linked Sonas to cloud storage. He believes that in the current storage situation, on the one hand, a large number of unstructured data continues to grow, the requirements for storage systems are increasing; on the other hand, large data access to cloud storage also requires storage systems. In addition, other vendors, including Symantec and Isilon, have also referred to the cloud storage level for cluster NAS applications.

Cluster NAS should be one of the most reliable application options when all kinds of products are related to cloud computing and cloud storage.

Let's take a sonas example to look at its architecture model in the cloud storage environment. The architecture of different vendors may vary, but not too much.

As you can see from the figure above, the Sonas uses a Infinband network for internal data interchange, which includes the interface node, the Storage node, and the management node three processing nodes, which are responsible for data exchange, data storage and data management respectively. In this, Interface node undertakes the task of data exchange, with a minimum of 2 nodes and a maximum of 30.

As we've said before, one of the main components of cluster NAS is its file system. The Distributed file system used by IBM on Sonas is called GPFS, and users can manipulate data through common file access protocols and TSM archiving software interfaces. At present, in addition to sonas this kind of integration file system, x86 hardware and network equipment products, there are simple file system products, such as Symantec's Filestore.

Because of the advantages of the system, the number of files that can be stored in the cluster NAS system is more than billion, for the actual application, the number of files limit can be negligible. At the same time, the global namespace is also the size of several PB above, some manufacturers of global space capacity can even reach 10PB.

Currently, all vendor cluster NAS devices can make a perfect transition to the original system, and users can actually transfer data to a clustered NAS system only if they need to migrate locally or remotely. This handy new and old transformation also allows many CIOs to start thinking about using clustered NAS products when building private clouds. It can be said that the discovery of the cloud storage market, to expand the cluster NAS a wider application area, which makes more and more manufacturers start to launch the corresponding products.