Finally a reality

The idea behind tiered storage is not new to anyone in the IT game, and certainly not to those who spend their days discussing, designing and deploying ways of storing important data for the enterprise.

Tiered storage has been difficult to implement and even more complex to automate - and most certainly near impossible to police.

Adam Day is product manager at SYSDBA.

Tiered storage has been a goal for some time, and a slew of different iterations have been touted in the market by storage hardware and software vendors.

Tiered storage through the years has gone by various names - for example, HSM (hierarchical storage management) and ILM (information life cycle management) - as well as many other names dependant on which high priced marketing team was trying to get the messaging and product accepted by the marketplace.

Not that this is a major problem; those in the IT industry are always looking for new acronyms or ways of marketing and repackaging good concepts. Perhaps it is believed that if a technology has a new moniker, the financial manager won't query why he or she is being asked to repeatedly purchase the same solution.

However, failed attempts aside, the desired result of tiered offerings most definitely justifies the effort.

The goal is to store data in an appropriate location, determined by relevance and business value. So, for example, the MP3 and video repository should never reside on the same solid state tier invested in to make the database perform like a Ferrari, although I am sure employees would appreciate the performance when watching videos over the costly infrastructure.

The problem is tiered storage has been difficult to implement and even more complex to automate - and most certainly near impossible to police.

Some of the reasons behind failures in the implementation of tiered storage are that the products have not been significantly granular to have a large reduction on cost, nor have they been sufficiently automated to provide the benefits of simplicity and automation to the business.

In reality, it has just been too costly and complex to deploy and business has not be unable to justify the costs relating to implementing the solution over simply purchasing additional hardware.

However, the goals of tiered storage, outlined below, are intriguing because they offer some simple, yet major business benefits:

* Price - Reduce the cost of storing data by ensuring it is on the relevant platform as not all data carries the same value to an organisation.
* Performance - Enterprise applications have different performance requirements, not all information requires the performance that solid state disk offers.
* Capacity - This goes hand-in-hand with performance and price. For example, large capacity databases often require both performance and capacity. These platforms entail complex architectures to perform at peak - and most databases have the majority of data accessed infrequently, the solution invariably becomes costly due to lack of credible information or skills, so problems end up being resolved by hosting the entire data store on a costly storage tier.
* Function - This has various aspects, but most commonly focuses on the availability and compliance aspects of storage.

It seems that these goals are finally - and rapidly - becoming reality. Nearly all major vendors are making progress in the implementation of automated data tiering technologies and have, at a minimum, identified the following needs:

1. Tiered environments must to be simple to deploy.
2. Tiered disk must be achieved at a granular level (sub lun), allowing single application environments to span multiple storage tiers with data movement between tiers.
3. This process must be automated and seamless to the enterprise.

There have been a number of announcements in the media recently about the advances in these technologies; keep an eye on the specifics of the vendor products in terms of progress in real world implementations of tiered storage solutions against the messaging being delivered.

If companies are in the market for a storage solution or looking to refresh their data store, they now have access to technologies that can assist in automating data tiering while allowing them to get the most out of their investment, be it in a solid state layer or making sure the data resides on the appropriate layer.

With these advances in mind, don't approach buying an SSD device or archiving solution as might have been done in the past, by simply comparing terabytes to terabytes or, when needing performance, IOPs to IOPs.

Rather investigate the advances being made by storage solutions in the automation of data movement across the tiers; companies may just end up saving themselves a large chunk of their budget, while still achieving the extremely difficult business and IT goals of reducing the cost and complexity of the platform - while in parallel achieving performance goals.