In the server application space, the capability exists to spread the storage around different types of disk. That's not to say that the same file, i.e., a SQL Server or an Exchange database, can't be spread over different disk types (with products like
Pay attention to how many disks each type of application has at its disposal. It is important to provision sufficient disks to meet the IOPs requirement. It is no good having different classes of disk if the performance of your application suffers because you didn't provide it with the sufficient, appropriate storage.
At the SMB level the options are fairly limited. As businesses move upward in the amount of data they have, the likelihood grows that there will be a budget available for hardware or software measures to protect and maximize their investment. While databases can be purged through application-side policies and archiving, by far the most problematic issue is unstructured data.
Microsoft Word, Excel, PowerPoint and similar files are difficult to move without impacting the ability to quickly locate them in the future. While dealing with this in a large enterprise rather expensive, it is still quite simple. Deploying a product such as F5's Acopia ARX system or Virtual File Manager from NetApp, will help spread the unstructured files around disk types with no impact on the directory location of that information, generally referred to as a single or global name space. With a smaller business it is a challenge, even though the numbers of files are of an order of magnitude fewer.
An organization without deep pockets and a fixed set of clearly defined requirements may want to look at a product such as the Compellent Storage Center. This product allows storage administrators to set aside different types of disks (Fibre Channel, SATA, SAS), create a single LUN that is able to span the different disks and to assign policies based on the change rate of the disks blocks within that LUN.
It may be that all new data is written to the faster and thereby generally more expensive disks. After a certain period of time, if the blocks on the disk have not been changed, the data is rewritten to slower disks. This frees up space for new data that is likely to be referenced in the immediate future. While the files are still immediately available in the same directory structure that they always have been, they are now on disks that might be a terabyte in size and not of a particularly high performance level.
SMBs with greater financial resources and a different magnitude of storage problems should use products such as NetApp VFM to provide that single namespace. Data is physically moved from one array to the other, seamlessly. In addition to reallocating files to storage based on their age, moving files based on their types to other storage media is also of interest to many companies.
Many organizations have problems with their staff using the network storage for personal media such as MP3 music, JPEG images and other file types that can be easily categorized. Moving artwork, logos, commercial video and such from the expensive fibre channel storage to high capacity SATA storage can save a lot of money. It also allows an organization to run with a fixed amount of Fibre Channel disks and only have to worry about planning progressive capacity upgrades to the tier two storage.
Using storage tiers has always required careful planning to ensure that the right amount of storage is available and that users can always find the information they need, preferably where they left it. There is no single answer for an organization. Entering into a tiered storage project should be undertaken with the assistance of your storage vendor who should provide the most economic and flexible options available to you.
About the author: Mark Arnold, MCSE+M, Microsoft MVP, is the principal consultant with LMA Consulting LLC, a Philadelphia, PA-based private messaging and storage consultancy. Mark assists customers in designs of SAN-based Exchange implementations. You can contact him at email@example.com.
This was first published in July 2008