Tiered storage often means the development of a storage strategy that moves data from fast, expensive storage to slower SATA based storage. Tiered storage can also refer to the quality of the storage. A tier 1 storage system may have a higher level of reliability and DR functionality than a tier 2 system. The delta between the quality and capabilities of these tiers is quickly eroding, which makes the question a fair one to ask.When using "tier" to refer to the quality and capability of the system the focus often centers on a tier 1 storage platform. This is most often thought of in terms of large monolithic systems traditionally supplied by EMC and HDS. These systems often build out one frame at a time with drives being added to the frame as capacity is needed. The systems often have advanced software and backend storage interconnects to offer extremely high performance and reliability. Their use is often confined to data centers with extreme uptime requirements and performance needs.
Tier 2 systems offered by companies like NetApp, Xiotech, Compellent and Nexsan are typically represented by more modular systems where capacity is added to a storage compute engine one shelf at a time. These shelves typically have the drives pre-installed. These systems often have moderate storage software with moderate storage interconnects. The scaling and performance of these architectures is significantly greater than it used to be. As a result tier 2 systems often meet the storage I/O needs of many data centers. Eventually however there can be a limit when scaling. At some point capacity or I/O demands could outstrip the capabilities of the storage compute engine. This used to mean either the addition of an additional storage system or the upgrade to a tier 1 storage platform.
Blurring the line is systems with more clustered architectures like those from 3PAR, Isilon and Pillar Data. Even EMC has jumped into more of a clustered storage approach with their VMAX architectures. Clustered storage systems address the limitations of modular and eventually even monolithic systems by offering greater scale and performance. Most of these systems can grow performance and capacity independently. The result is that sophisticated scalable architectures are no longer the sole domain of tier 1 storage.
If performance and scale are no longer the sole property of tier 1 storage the remaining differentiation then becomes how does the software differentiate itself. The expectation of tier 1 storage systems is that they will offer non-disruptive upgradability, a resilient architecture that can maintain performance during failure and a multi-site, multi-mode disaster recovery capability.
For example the capability to replicate data synchronously (mirrored) to a second location miles away, then from that second location replicate (asynchronously) to a third location hundreds of miles away. The challenge is first that not every data center needs this kind of reliability and if they do companies like 3PAR, NetApp and others are closing the gap on the software side to bring this type of advanced functionality to the other storage tiers.
When tier one storage vendors start using standard hardware and more common interconnect architectures like clustering, the focus centers on the quality of the storage software. If the software capabilities of what is traditionally considered tier 2 or clustered storage closes the gap between its tier 1 software counterparts, you are left wondering why use tier 1 storage at all or at least including more vendors in the tier 1 designation.
Track us on Twitter: http://twitter.com/storageswiss
Subscribe to our RSS feed.
George Crump is lead analyst of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. Find Storage Switzerland's disclosure statement here.