Real-World Speed Tests for Different SSD RAID Levels

This tutorial measures and illustrates the difference in speeds between different RAID levels for SSD (solid state drive) RAIDs. This also provides suggestions on which RAID level to choose for your storage.

NOTE: Here’s a similar tutorial covering HDD RAIDs.

RAIDs can be formatted into different RAID “levels.” For example, RAID 0, 1, 4, 5, 6 and 1+0. Which you choose balances multiple factors:

• Whether the RAID uses SSD or HDD drives
• The number of drives
• The speed you want
• Whether you want automatic data backup
• Whether you want your data protected in case a drive dies

The more protection, the slower the speed and the lower the storage capacity. There’s no “right” answer, just the answer that works best for you. Like most of tech, each choice is a tradeoff.

NOTE: Here’s an article that describes each RAID level in a bit more detail.

Recently, I purchased an OWC Thunderblade, a 4-blade NVMe SSD RAID. (A “blade” is the SSD equivalent of a HDD drive.)  Along with this, I purchased SoftRAID, RAID controller software published by OWC, and four Seagate Iron Wolf Pro 16 TB hard drives. This system can be configured as RAID 0 or 1 using Apple Disk Utility, or RAID 0, 1, 4, 5, 1+0 using SoftRAID. For these tests, I used SoftRAID v.7.

The Tunderblade was connected via Thunderbolt 3/4.

Here are the results. Results are displayed as a percentage because actual speeds will vary based upon the number of blades, how they are connected, the speed of your computer and a few other factors. RAIDs with more blades will be faster, but the speed relationship between different levels remains the same.

NOTE: RAID levels can only be set or changed when the RAID is empty. If you need to change the RAID level, first copy all the data you want to keep to another location.

NOTE: These tests were made with an OWC ThunderBlade NVMe SSD RAID formatted using SoftRAID and tested on an M1 Pro MacBook Pro.

Reads are always faster than writes. And, while none of the numbers perfectly matched the theory, they were all reasonably close.

• RAID 0 is the fastest because all four blades are used for writing or reading data. This provides 100% of the total storage capacity of the RAID, but no protection in case a blade dies.
• RAID 1 is the slowest because it is always, regardless of the number of blades, the speed of a single blades. As well, this only provides the storage capacity of a single blades regardless of how many drives are in the RAID.
• RAID 4 reserves one blades for “parity data” which reduces overall speed. But if one blades dies, all your data is protected. RAID 4 is optimized for SSD RAIDs. This provides 75% of the total storage capacity of the RAID for a 4-blades RAID.
• RAID 5 shares parity data across all blades, also reducing overall speed. But if one blades dies, all your data is protected. RAID 5 is optimized for HDD RAIDs. This provides 75% of the total storage capacity of the RAID for a 4-blades RAID.
• RAID 1+0 reserves half your total blades (2, in the case of a 4-blades RAID) for parity data. It is optimized for data recovery with excellent speed. This provides 50% of the total storage capacity of the RAID. This is ideal when you want two identical copies of the data on your RAID.

NOTE: Protecting against drive failure is not like Time Machine or backup software. There are no versions or multiple copies of the same file. Hardware protection means that if a drive dies, existing data is safe. If you erase a file, hardware protection will not help you recover it.

SUMMARY

For SSD RAIDs, choose RAID 0 for speed, because SSDs are generally more reliable than spinning media. Then, to be safe, backup your data every night. If you still want to protect against drive failure, choose RAID 4.

EXTRA CREDIT

• Here’s a PDF containing my SSD test data and testing notes.
• This article details how the tests were performed, which was part of a detailed product review of the Thunderblade.

---

Bookmark the permalink.