Q
Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

What is the best SSD for server storage improvement?

Should data centers use SAS- or PCIe-based on-server flash devices? What options are out there for SSDs on servers?

As flash memory's costs decline and long-term reliability improves, SSD flash storage is finally seeing volume...

deployments in data centers. PCIe-based flash storage may face fewer bottlenecks inside data center servers than SAS-based devices, but multiple options are available.

Flash storage components – solid-state drives (SSDs), solid-state accelerators (SSAs) and storage-on-DIMM devices -- are proliferating in mainstream and niche data center server scenarios.

SSDs have been around for years, applying a large volume of flash memory behind a standard serial-attached SCSI (SAS) interface. This allows operating systems and firmware to interact with an SSD just like they would with a magnetic hard disk drive (HDD). An SSD for server installation boosts local storage capacity or handles tasks such as page swapping or file caching. SSDs also appear in enterprise storage arrays as a high-performance media with HDDs or all SSDs.

Commonly available SAS-1 and -2 interface speeds of 3 and 6 Gbps pose a potential bottleneck for SSD storage traffic. Data centers are adopting servers and storage with 12 Gbps SAS-3 interfaces, and a 24 Gbps version dubbed SAS-4 is in development. But SAS is still slower than a server's internal PCIe interface, at over 15.7 Gbps per lane with multiple lanes aggregated when necessary. A PCIe interface using four available lanes offers about 63 Gbps of bandwidth.

An emerging trend is to use PCIe-based flash memory SSAs, also called I/O accelerators, as local high-performance storage on servers. An SSA comprises a large SSD that communicates over PCIe instead of SAS. Accelerators suit all types of demanding storage-related tasks, such as serving databases for demanding big data projects, holding VM or virtual desktop images, and storing data for large enterprise applications like Exchange.

A relatively new SSD for server use lets data centers install flash storage devices next to the processor on available DIMM slots. Storage-on-DIMM, also called flash-on-DIMM, devices reside within close proximity to the processor, yielding extremely small read and write latencies. IOPS can exceed those of similar SAS- or PCIe-based flash storage devices. Nonvolatile storage-on-DIMM devices suit the same range of tasks as SSAs.

The best SSD for server-based storage isn't just the one with the fastest reads or writes. Evaluate what performance you get for the price you want to spend, multiplied by the number of servers in the data center that need SSD, to determine the right product.

About the author:
Stephen J. Bigelow is senior technology writer in the Data Center and Virtualization group at TechTarget. He has written more than 15 feature books on computer troubleshooting. Find him on Twitter @Stephen_Bigelow.

Next Steps

It's time to start using SSD storage

Role of PCIe cards up for debate

How a storage accelerator improves access time and IOPS

This was last published in September 2015

Dig Deeper on Enterprise data storage strategies

PRO+

Content

Find more PRO+ content and other member only offers, here.

Have a question for an expert?

Please add a title for your question

Get answers from a TechTarget expert on whatever's puzzling you.

You will be able to add details on the next page.

Join the conversation

1 comment

Send me notifications when other members comment.

By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

Please create a username to comment.

What do you think is the best SSD for server storage? Is there a setup that you prefer in your data center?
Cancel

-ADS BY GOOGLE

SearchWindowsServer

SearchEnterpriseLinux

SearchServerVirtualization

SearchCloudComputing

Close