NAS Caveats

The caveats for NAS are twofold. The architecture employed by NAS products does not lend itself to applications that are dependent on sophisticated manipulation of the storage media—meaning any application that either has its own embedded file system to map and store data to the online storage directly, or that works with data management products, such as relational databases, which do the same. Secondly, applications that deal with highly volatile data do not perform well—in other words, data that is subject to constant and continuous updating by a large user base.

Writing to disk directly is a characteristic of many applications first developed within the UNIX operating environment. Relational databases commonly have this architecture. The problem manifests itself in performance as these systems have developed their own ways of manipulating data and in a sense have their own internal file system. Having these run under an additional file system penalizes the application by forcing it to do certain things twice.

However, any application that contains a database designed in a monolithic fashion will have performance problems when operating within NAS storage environments. For example, e-mail applications have embedded databases that are not scalable. The monolithic nature of their databases can therefore force all traffic into a single path or device creating a performance bottleneck and recovery problem if the storage device or path is encumbered in any way.

On the other hand, NAS provides the most cost-effective solution for applications that are file-oriented, contain read-only files, or which have large user bases. The architecture behind NAS within these application environments provides a scalable way for large numbers of users to obtain information quickly without encumbering the servers. This also provides an effective way to upgrade storage quickly through existing networking resources. This has proven itself within the ISP data centers where NAS storage devices are the solution of choice given their capability to handle large amounts of data through the storage of web pages, and users who require random read-only access.