Tiered storage is a storage method which involves storing data on a range of different storage media with different characteristics, such as performance, cost, and capacity.
The different storage media are organized into a hierarchy, where the highest performance storage media is deemed to be Tier 0 or Tier 1, followed by Tier 2, Tier 3, and so on.
Tier 0 or Tier 1 is usually made up of flash or 3D XPoint-based solid state drives, while successive tiered storage levels may involve high performance fibre channel or SAS drives (or RAID arrays containing them), lower performance SATA drives, optical disks, tape storage systems, and cloud-based nearline or offline storage systems.
A 3 tiered storage model using disks and tape to provide tier 1, tier 2 and tier 3 storage was perhaps the most popular tiered storage model before SSDs and cloud storage became commonplace.
But now it is not unusual to use a tiered storage model that involves five or more tiered storage levels. Each one is subtly nuanced to produce a different combination of the three key storage attributes: cost, performance, and capacity.
If expense were no object, businesses would use SSDs for all their storage requirements because they offer very high performance and reliability.
But in the real world storage costs are very important, as IT departments seek to live within their budgets and organizations as a whole seek to minimize costs and maximize efficiency. And unfortunately SSD storage is expensive compared to hard disk drives, and far more expensive than tape storage.
That means that SSDs need to be used judiciously, and only for storing data that is used by systems that require very high performance. Data that is less important can be stored on lower cost, lower performance systems such as hard drives, and data that is rarely accessed, or is only retained for compliance purposes, can be relegated to very low cost offline storage systems
So tiered storage is not driven by the needs of the IT department for operational reasons, but purely for financial reasons. The objective of a tiered storage system then is to minimize storage costs by using the least costly storage option possible which provides the minimum required performance.
A tiered storage system with just two storage tiers offers storage administrators a very limited choice over where a given piece of data should be stored. If a certain level of storage performance is required for that data then it will be stored in Tier 2 if that offers the required level of performance, or, if it does not, then it will be stored in the more expensive Tier 1.
A tiered storage system with three storage tiers is likely to be more efficient, because that same piece of data can be stored in Tier 3 if that offers the required level of performance, or if not then it can be stored in Tier 2 if that offers the required level of performance, and only when both Tier 2 and Tier 3 fail to offer the required level of performance does it need to be stored in the most expensive Tier 1 storage.
In fact the more storage tiers there are available, the more efficiently a piece of data can be stored in an appropriate storage tier that meets its performance needs, rather than one that which offers unnecessarily high performance at unnecessarily high cost. New tiers can be created from a small set of different storage media types using techniques such as creating RAID arrays or short stroking disks. Both of these techniques, for example, create higher performance storage at the cost of lower storage efficiency (and thus, ultimately, higher financial cost).
The biggest challenges presented by a tiered storage system are classifying data into multiple classes, deciding which storage tier best suits a given class of data, and reclassifying data on a regular basis as it ages.
The key point here is that any given piece of data’s storage requirements may change over time, so rather than storing data in a tiered storage system and forgetting about it, the data has to be monitored and moved to a lower (and lower cost) storage tier as soon as it no longer requires the performance of the storage tier that it currently finds itself in.
Typical data classes for tiered storage include:
Apart from the very smallest businesses, it is clear that a manual approach to storage tiering is likely to be too time consuming and cumbersome to be successful. For that reason, most tiered storage systems rely on automated storage tiering using a tier management system which monitors data throughout its lifecycle, and moves it down the storage tiers automatically as it cools.
This software may be supplied with, and operate on, a single storage system, or it might be a standalone solution which works right across an organization’ storage infrastructure, and in some cases out into the cloud as well via a cloud storage gateway.
In some organizations which have used storage tiering for some time, new, faster storage technology than used in their existing Tier 1 has become available. For that reason, this technology is adopted as “Tier 0” storage. But for the purposes of this article Tier 1 will be assumed to mean the highest performance storage tier.
This storage tier is designed to store data which is highly volatile and time-sensitive and which needs to be available in the shortest possible time. Typically used in financial trading environments or other business areas where far larger amounts of money than the cost of the storage can be made or lost in a fraction of a second, only the fastest possible storage is sufficient. For that reason very fast solid state storage media configured for the highest possible performance, regardless of overall storage efficiency, is usually required for Tier 1 storage.
This storage tier is used to store transactional data to support high-performance applications, customer-facing systems such as retail applications, and other systems where anything more than extremely short delays are unacceptable. Since the performance levels required are not quite as high as for Tier 1 purposes, lower cost and more storage efficient solid state storage systems are typically used to provide a Tier 2 storage solution.
The third storage tier is used to store “hot” data such as CRM and ERP data, and even recent emails, which needs to be accessed frequently without undue delay. That means that it needs to be stored on media such as medium to high performance hard drives which are relatively low cost compared to solid state storage media. In many organizations there will be substantially more Tier 3 data than Tier 2 or Tier 1, so high capacity hard drives which offer a low cost per Gigabyte stored provide an ideal solution.
This storage tier is used to for “warm” data such as data from recently completed transactions, emails which are more than a few days old, and other data which will be accessed infrequently. It can also be used to store data which may be used for business information and data visualization system, or financial data which will be needed for creating monthly, quarterly or annual reports. but nevertheless needs to be accessible relatively quickly whenever required. Tier 4 storage requirements can be very large, and since cost is a major consideration a typical solution for Tier 4 storage is high capacity hard disk drive storage. Rather than high performance SAS disks or RAID arrays configured for performance, Tier 4 storage is more likely to use low cost, relatively low performance SATA drives.
The lowest storage tier is typically used for archiving “cold” data which may never be accessed again, but which still retains some value (perhaps for data mining in the future) and which is therefore worth retaining. It is also used to store data which must be retained for compliance purposes, and which only needs to be accessible within days or weeks rather than minutes or seconds.
Optical media or tape storage systems which offer storage at a very low cost per Gigabyte, but with very low performance, are typically used for this storage tier. In the last few years cloud-based storage such as Amazon’s Glacier storage service has become popular for companies who wish to offload large amounts of data for low cost storage with the possibility or retrieving it relatively easily in a matter of hours if ever required.
Tiered storage brings a number of benefits, the most important of which are:
Enterprise Storage Forum offers practical information on data storage and protection from several different perspectives: hardware, software, on-premises services and cloud services. It also includes storage security and deep looks into various storage technologies, including object storage and modern parallel file systems. ESF is an ideal website for enterprise storage admins, CTOs and storage architects to reference in order to stay informed about the latest products, services and trends in the storage industry.
Advertise with TechnologyAdvice on Enterprise Storage Forum and our other IT-focused platforms.
Property of TechnologyAdvice.
© 2021 TechnologyAdvice. All Rights Reserved
Advertiser Disclosure: Some of the products that appear on this site are from companies from which TechnologyAdvice receives compensation. This compensation may impact how and where products appear on this site including, for example, the order in which they appear. TechnologyAdvice does not include all companies or all types of products available in the marketplace.