What you will learn: This tip discusses the pros and cons of a tiered SAN architecture, as well as the factors to consider before creating a tiered SAN.
The concept of tiered storage began with storage devices, but has quickly spread to the storage area networks (SANs) those devices are connected to. Tiered SANs can offer even greater savings than tiering storage devices, but with these savings comes a more complex architecture and a need for increased management.
((Content component not found.))
The concept behind a tiered SAN architecture is simple. At the top of a tiered SAN infrastructure might be a fast Fibre Channel SAN, with high levels of redundancy and disaster recovery features. This Fibre Channel SAN would handle applications for which speed and availability are critical. The second tier would be a lower performance SAN with perhaps only limited disaster recovery features. This SAN might rely on less expensive SATA disks and less elaborate data protection schemes with longer recovery times and slower connections. The availability of Fibre Channel alternatives, such as iSCSI, can make tiering even more attractive.
A tiered SAN architecture can also allow the testing of new SAN technologies without putting critical applications at risk. A lower-level SAN can provide a more extensive test than the usual prototype installation.
Here are three things to consider when planning a tiered SAN architecture:
How to design your tiered storage system
Which functions will each SAN support? A good start to figuring this out is by classifying the supporting applications according to the necessary performance characteristics. Do you need full disaster recovery? What is the preferred access time over the SAN? What is the best topology for your needs? Do you want to incorporate remote sites into your tiered SANs? What about QoS agreements?
Perform the basic economic calculations
Will a tiered SAN architecture save you enough money to make it worthwhile? The answer isn't always yes, especially with small or medium-sized SANs. There may not be enough applications in a given class to make it worthwhile to establish a new SAN.
If the projected costs are close, you may be better off expanding your existing SAN, since such projections are notoriously fuzzy. If a single technical glitch can eat up all your capital savings, tiering may not be your best choice.
Decide on management options
Tiered SANs are inherently more complex to manage. If you design your SANs for unified management, this may not be much of a problem, but it is something to think about. Will your present SAN management tools handle your new SAN architecture? Try to avoid situations where you need different tools to manage each of your SANs.
About the author: Rick Cook writes about issues related to storage and storage management.
