Memory & Storage

Improve Your Storage Architecture Performance With Automated Storage Tiering

Typically, storage architecture configurations include storage systems with multiple tiers of storage, each tier with more speed, less latency and less capacity than the one below, since costs go up with each level of performance. Automated storage tiering ensures that the most-used data is kept on the fastest available storage — data that hasn’t been accessed for a set interval is moved to a slower, less expensive tier, while data that was on a lower tier is moved up when someone accesses it. This all happens transparently to the end user and even to the administrator, with files in the same directory tree stored on different tiers depending on how often the data is accessed.

Automated storage tiering as a SAN technology has been around for over a decade, initially with several tiers of hard drives: 7,200 RPM, 10,000 RPM and 15,000 RPM drives. Each offered more speed at a higher cost than the one before. A typical system might have 15 300 GB, 15,000 RPM SAS drives for Tier 1; 15 600 GB, 10,000 SAS drives for Tier 2; and 15 1.2 TB SATA drives for Tier 3.

When storage architectures featuring the first enterprise-class flash drives became available, they were added to the top of the pyramid as Tier 0, with the fastest performance and the highest cost per gigabyte. In 2009, with high-speed SSDs costing over $20,000 and having capacities of only 147 GB, it was challenging to write algorithms to make the best use of the very limited space at the top of the pyramid, and a typical system might have only three to five 147 GB SSDs making up Tier 0.

Discover the Cost Advantages of SSDs

White Paper

Read this white paper to learn the total cost of ownership advantages of using SSDs. Download Now

Over the last eight years or so, SSD capacities have gone from the meager 147 GB to 16 TB, while prices in dollars per gigabyte have dropped from the $136/GB of the 147 GB drives to $0.625/GB for a $10,000 16 TB SSD available today. With even lower prices per GB for smaller SSDs, and the added effects of compression and deduplication, SSDs can reach an effective price-per-GB comparable to hard disks in SAN systems.

Increased Levels of SSD Performance

At the same time that prices have dropped, the rapid pace of improvements in SSD technologies has resulted in several tiers of performance for SSDs, comparable to the 7,200, 10,000 and 15,000 RPM hard drives of the earlier systems. With the advent of NVMe PCIe SSDs, such as the Samsung 960 Pro NVMe, performance has improved substantially, with 5x improvements in speed and substantial latency reductions over earlier SSDs.

Additionally, enterprise-class SSDs like the Samsung SM863a and PM863a have been developed that have life spans much longer than earlier drives, along with exceptional performance. So, a three-tier SSD solution is now not only possible, but useful, with an ultra-fast NVMe tier, an enterprise-grade read/write tier and a read-intensive consumer grade tier. Each offers higher performance than the one below, with higher costs as well.

Advantages Over Caching

A storage architecture built on Tier 0 and one built on caching have similar goals — use a faster type of storage to accelerate the input/output operations for the entire storage system. However, cache is a temporary storage area that data is written to, and then sent along to the final resting place for the data. This might be done with a block of RAM on the server and a SAN system, or with a fast server-side flash device and SAN, or with an appliance that is placed between the server and the storage system.

In any case, if data is lost after being written to the cache but before it’s written to the storage system, the server won’t know that data has been lost, which makes recovery more difficult. While cache can be backed up with batteries, and algorithms established to ensure confirmation of data being written to permanent storage, these also eat into the advantage that cache offers in acceleration. Tier 0 offers much of the same advantages of caching, but a main difference is that data is always written to permanent storage, so there is less chance of data loss or missed synchronization of the data between server and storage.

Should You Use It for Your Storage Architecture?

While automated storage tiering is typical of high-end SAN systems, it’s not solely the domain of these systems. It’s entirely possible to use data management software to create automated tiering with existing storage systems, or with drives installed within a single server. The process of tiering is worthwhile for administrators to try with free demo systems, as it can be beneficial for the performance advantages it offers without substantially increasing costs.

Check out our full line of SSDs to find the perfect storage solution to address your organization’s needs.

Posts By

Logan Harbaugh

Logan Harbaugh is an IT consultant and reviewer. He has worked in IT for over 20 years, and was a senior contributing editor with InfoWorld Labs as well as a senior technology editor at Information Week Labs. He has written reviews of enterprise IT products including storage, network switches, operating systems, and more for many publications and websites, including Storage Magazine, TechTarget.com, StateTech, Information Week, PC Magazine and Internet.com. He is the author of two books on network troubleshooting.

View more posts by Logan Harbaugh