Hybrid storage arrays combine the high performance of SSDs with the high capacity and low cost of HDDs. Although...
these hybrid storage systems include the same types of features as the high-end all-flash arrays, they also offer automated tiering to optimize use of the flash part of the array. As you get ready to examine the leading hybrid storage arrays, it is important to understand the features and functionality this technology offers and how this could benefit your organization.
Unlike older technologies, the latest hybrid storage systems essentially virtualize storage. The administrator does not need to create specific volumes for different performance levels. Instead, the storage system simply identifies the most-used files and places them on the fastest available tier of storage. When those files are accessed less frequently, the system moves them to less-expensive tiers.
The emergence of flash brings more tiers to enterprise storage arrays. Potential tiers include memory bus flash; nonvolatile memory express (NVMe) flash; high-performance SATA flash; low-cost SATA flash; 15,000, 10,000, 7,200 and 5,400 rpm HDDs; tape, RDX removable disks; and even the cloud. The speed of the fastest SSD tiers enables other storage features, such as inline compression and deduplication, which can increase the effective capacity of an array by as much as five or six times in real time.
Autotiering was originally developed to keep the fastest tiers of HDD storage fully utilized. Now, the same technology can coordinate several different types of storage, not only on different types of HDD or flash, but in multiple arrays or even across data centers. Each tier increases in speed and cost over the tier below. There is no need to use every possible tier in a system, however. One high-performance SSD tier, one low-cost, high-capacity HDD tier and an archive or cloud tier will provide excellent performance and more capacity than an all-flash array at a given price point.
Autotiering software can also take into consideration the cost of the storage, the distance to the server using the data across multiple data centers on premises or in the cloud and the speed requirements at specific times. For example, you may need faster performance to support end-of-month reports or anticipated peaks in demand for e-commerce or content delivery.
Features such as deduplication and compression were originally used only on offline volumes because of their impact on performance. However, a portion of the flash tier of a hybrid array can provide a landing zone to deliver these features, even for the HDD portions of the array.
Using extensive research into the hybrid flash storage market, TechTarget editors focused on market-leading vendors and other well-established enterprise vendors. Our research included data from TechTarget surveys, as well as reports from other respected research firms, including Gartner.
Considerations for purchasing hybrid storage systems
Cost. One of the main reasons for buying hybrid storage is its lower overall cost over all-flash arrays. Even with a five or six times effective compression ratio using deduplication and compression, high-capacity HDDs can provide lower-cost storage for less-active and inactive data without having to send it off-site to a tape or cloud archive. But off-site storage systems can be considerably less expensive than the HDD tier for data that must be kept but will not need to be accessed. Tiering software that can move the data automatically to an HDD, cloud or tape tier will reduce costs substantially, especially for archived data.
Ease of use. Some hybrid storage systems are harder to use than others. For the CTO or data center admin who might have to configure storage when a trained admin is not available, a system with an interface that can be figured out without a week of training offers an advantage.
Single-pane management of multiple systems. Along with ease of management of the system itself, some hybrid storage systems enable the administrator to connect other types of storage. Therefore, an admin can integrate and manage a variety of products and services from a single console. These products include other disk or flash storage; tape archives or other removable storage; and off-site public or private cloud storage. Third-party management tools can also provide this type of functionality if the hybrid storage vendor is not capable of integrating multiple types of storage.
As vendors strive to differentiate themselves from their competitors in a crowded storage landscape, they continually come up with new features. These include introducing cloud storage gateways and integrating existing legacy storage systems into their controllers. There are dozens of unique features offered in this space, and major features are trickling down from high-end million-dollar systems to $500 NAS boxes.
Multilevel hybrid arrays
The term multilevel takes on new meaning given the extended tiers, types of flash, types of HDD, removable storage and hot, warm and cold cloud tiers. A dozen tiers could be integrated into a single system. In practice, most admins would find three or four tiers sufficient, depending on the overall requirements of the system, as well as the specific demands of the various applications using the storage. There might be an online transaction processing or real-time application that requires the fastest flash possible and other apps that are lower priority but that still need NVMe flash. Or there might be a larger app that is best suited for high-performance SATA SSDs and a large database that is already running on an HDD-based system, along with cloud apps that use multiple cloud tiers.
The proliferation of tiers can affect management of the storage. Using only the included management utilities for each separate type of storage, an administrator might need to use four or five separate apps to set up and run all their storage. More software will be required to migrate data from one system to another. Using a storage product that can handle external tiers can save substantial time and effort, especially over the life of the data.
Hybrid storage systems streamline storage operating costs
Data retention standards require keeping patient records or other data for extended periods of time. Over these types of retention periods, power usage becomes a major factor in the overall cost of storing the data. Long after the hardware has been amortized, the recurring operating costs remain significant. Data saved to tape or other removable media will not use power once removed, and cold storage options in cloud data may cost less than running existing hard drive storage systems. However, both SSD- and HDD-based storage systems may be able to run in power-saving modes. HDDs have both standby and sleep modes, which save power but increase the time to initially access data. Standby reduces the drive speed, reducing power consumption, but keeps it powered on. Sleep mode powers down the drive, requiring as long as 20 seconds to power it back up and access data. The usability of these modes depends on both the storage array and the drives in use. If archiving is a priority, ask the vendor if it supports a low-power configuration.
SSDs can provide similar functionality without the long delay to turn the drives back on. The lower power mode can reduce power consumption by a factor of 10, with a relatively short delay to power the drive back up after a period of inactivity.