Tiered storage definition
Tiered storage is a system or method for assigning data to various types of storage media based on a range of requirements for cost, availability, performance, and recovery. Several factors determine which type of media, such as cloud storage, solid state storage arrays, disk or tape, is ideal for storing a particular category of data. These factors include cost, data recoverability, and data availability requirements.
Tiered data storage is part of the foundation of information lifecycle management (ILM) and helps companies reduce total storage costs while ensuring compliance and performance. For example, data retained for regulatory needs can be archived to save money, while data needed for restoration in the event of corruption, data loss, failed data migration, or other critical failure can speed recovery if it is stored locally.
How does tiered storage work?
One the greatest challenges a tiered data storage system presents is the need to identify and classify data to determine which storage tier suits each class of data best. Reclassifying data routinely as it ages is another challenge inherent to this kind of system.
A two-tiered storage system offers a limited choice. Level of storage performance will be the determining factor that divides all data up into two massive categories, and anything that cannot be moved to Tier 2 must be stored in the more expensive Tier 1 based on the required performance levels.
This is why a three-tiered storage system is typically more efficient. Administrators can choose to store data at the most costly level only when the required level of performance demands it. In fact, more storage tiers in general means more efficient, performance-based data storage.
What are the data classes for tiered storage?
Typical tiered storage data classes include:
Mission critical data. This data must always be stored at the highest tier because it supports high-speed applications such as customer transactions. Inability to access this kind of mission critical data means poor performance, lost business, delays, and other problems that negatively impact profitability.
Hot data. This data class is in constant use, so it demands a relatively high level of tiered storage. Data like this runs email, and applications such as ERP or CRM—anything the organization’s everyday operations depend on. At this tier performance is still very important, but cost is also a factor.
Warm data. This data is typically a few days old, and includes data on completed transactions and emails from days past—things that are used less frequently, but still must remain readily accessible. At this level, cost becomes the more important consideration, subject to a threshold for minimum performance.
Cold data. This sort of data has to be retained, but it may never be accessed again. Perhaps there are regulatory reasons to keep it, or it might be unstructured data that has big data mining uses in the future. Cold data like this should be stored based on low cost, with access times of minutes or hours.
Except for very small businesses, manual storage tiering is almost always too impractical and time consuming to work well. Most tiered storage systems deploy automated storage tiering, moving data through the tiers automatically as it moves through its lifecycle, cooling. This kind of tier data management system software may be a standalone solution, part of a cloud storage gateway, or part of a single storage system.
What are the data center storage tiers?
The phrase “Tier 1 data storage” sometimes causes confusion, because increasingly, the top layer in the tiered storage hierarchy is being referred to as Tier 0, thanks to the rise of technologies such as storage class memory and flash. However, for the purposes of this article, Tier 1 storage will refer to the storage tier with the highest performance.
Tier 1 data storage is designed for data which is highly time-sensitive, volatile, and must be accessed quickly—in as close to real time as possible. For example, in a stock trading environment, where huge amounts of money can be lost in an instant, only the fastest Tier 1 storage will work. Therefore, regardless of overall storage efficiency, at the Tier 1 level the storage media is typically very fast, solid state storage configured for the best possible performance, without regard to cost.
Tier 2 data storage. Tier 2 is typically used to store transactional support data for customer-facing systems such as retail applications, and other high-performance applications where only extremely short delays will work. A step down from Tier 1, a Tier 2 storage solution will usually utilize a more cost efficient storage system.
Tier 3 data storage. Tier 3 is used for hot data such as ERP and CRM data which users must access often without too much delay. Compared to Tier 1 and Tier 2, this usually means medium to high performance, high capacity hard drives at a low cost per Gigabyte stored and substantially more data.
Tier 4 data storage is where warm data from older emails and recently completed transactions goes—anything that needs to be accessed less often, but still regularly and without too much delay. Typical Tier 4 storage requirements include very large capacity and affordability, so high capacity, relatively low performance hard disk drive storage, such as SATA drives rather than high performance RAID arrays or SAS disks, are a more common solution.
Tier 5 data storage is for archiving cold data for the future. Since there is time to retrieve this kind of data and cost is the overriding factor here, cloud storage tiers, optical media, or disk and tape storage systems are great choices.
What is tiered storage architecture?
Modern multi-tiered storage architectures may be simple or complex. Two tier and Three tier storage architectures still exist, and some systems include five or six tiers, including storage tiering in cloud platforms.
What is multi-tiered storage architecture?
In a multi-tiered storage architecture, the various storage media are organized hierarchically. The highest performance storage media is Tier 0 or Tier 1, with the other tiers, Tier 2, Tier 3, and so on, following. Today it is common to see tiered storage models with five or more tiered storage levels. That said, some experts anticipate fewer tiers, with the most valuable data being stored on flash arrays and less-valuable data being stored in the cloud. The common factor is that a well-designed tiered storage system will ensure that each tier has a different, useful blend of performance, cost, and capacity.
Typically, flash storage or 3D Xpoint-based solid state drives comprise Tier 0 or Tier 1. Tape storage systems, high performance SAS or fiber channel drives, RAID arrays containing those kinds of drives, optical disks, lower performance SATA drives, and cloud-based storage systems may all be part of lower tiered storage levels.
Tiered storage vs hierarchical storage management
Although tiered storage and hierarchical storage management (HSM) are sometimes confused, they are not exactly the same. HSM typically refers to a data movement system in which movement happens automatically to some kinds of media depending on how often they are used. Tiered storage is where an HSM system might store its data.
Advantages of tiered storage
The advantages of data tiering storage include:
Reduced storage costs. Tiered storage allows companies to store each class of data based on the minimum performance that it requires and the lowest cost storage that can handle those requirements. This in turn eliminates the problem of paying for unneeded high performance storage.
Increased storage efficiency. High performance disk storage systems are often inefficient. Tiered storage systems move less critical data to lower tiers, reducing demand on short-stroked and RAID storage and improving storage efficiency.
Reuseable storage equipment. Equipment that doesn’t perform well enough for Tier 1 can still be reused in a tiered storage system.
Improved disaster recovery. Tiered storage infrastructures can significantly reduce the costs associated with achieving a zero recovery time objective (RTO) and recovery point objective (RPO) by reducing the amount of data the DR system must recover during a disaster. RPO and RTO are among the most important parameters of a data protection or disaster recovery plan because classifying data correctly ensures the data is in the best possible location for recovery at all times.
Storage tier optimization
Each of the tiers of a storage system involves fluctuations in four major qualities: performance, cost, capacity, and function. The needs of the type of data determine which of these qualities control and how many tiers are necessary to the system. Performance and capacity are typically more critical for the higher tiers, while cost controls on the lower end.
Functional differences also shape tiered storage requirements. For example, the need to replicate data for security reasons, or the high-speed data restoration function of a hard disk drive array. Data that is stored for different reasons and will be subject to different functions may need two tiers with functions suited to those goals.
Tiering vs caching
The basic difference between tiering and caching rests on the way systems utilize faster storage and detect and accelerate frequently accessed data using algorithms. Caching makes copies of data that is frequently used, checking to see if a copy already exists and reads from slower storage. Tiering identifies data and then moves it for use, without placing it in a copy area.
Does Druva offer tiered storage solutions?
Druva’s intelligent storage tiering ensures end-to-end data protection driven by machine learning, policy management via one-click actions across all backup tiers, and easier management through a central dashboard to manage long-term data retention, defensible deletion and governance processes.
Druva offers automatic tiering for long-term data retention of backup data, reducing our customers storage spend while ensuring compliance with data regulations.