Mastering NVMe Hot-plug: Navigating Challenges and Ensuring Safe Removal

Hot-Plugging, or the ability to insert and remove a device without having to power down one’s system, has been an important feature for users who are attempting to maximize their total uptime on often times mission critical systems. For decades this has been a standard feature supported by a wide variety of data storage devices like SAS/SATA Hard Drives (HDDs) and Solid-State Drives (SSDs). In addition, this feature is also a mainstay with most USB devices. Given that this feature has been around and is widely adopted by many different protocols, one might be forgiven in thinking that it would be standard with NVMe storage, however that is not the case. Unlike the SAS/SATA interface, which was designed with hot plugging in mind, NVMe instead, was designed originally to bypass the performance limitations of the aforementioned interfaces.

That is not to say that that NVMe media doesn’t have this feature at all. While NVMe hot-plug support isn’t as standardized as it is for SAS/SATA drives, there are some NVMe SSDs that do have the ability to hot-plug. However, this is no simple task; and this support is contingent on the following main factors. It is important to note, that all of these factors must be met; and if one is missing, NVMe hot-plug will not be supported.

1. System Requirements for NVMe Hot-plug: The system/motherboard the drives will be connected to must supports NVMe hot-plug. Typically, this information can be found with the associated user/technical manuals.

2. Operating System Support for NVMe Hot-plug: Certain operating systems and drivers are designed to be able to handle hot plugging of NVMe devices.

3. Drive Requirements for NVMe Hot-plug: Hot Plugging is a feature typically supported by DC or Enterprise grade NVMe drives. As such this is a feature supported by U.2/U.3 NVMe drives. M.2 drives, were not designed to support hot-plugging and since most of them are client grade, they don’t have the support to begin with.

4. PCIe Switch Solution for NVMe Hot-plug: NVMe drives require dedicated amounts of resources. A PCIe Switch will ensure that system CPU provides the resources to each channel where an NVMe could go.

Assuming you meet the criteria of all three of the requirements, this brings up the initial question, is it safe to simply remove the NVMe SSD from the system and replace it with a new one? In short, no. While it is possible to remove and replace an NVMe SSD from a running system, there are a series of best practices for NVMe Hot-Plugging that need to be taken in order to avoid any issues. Primarily this requires understanding of both the device and host side. In this scenario, the NVMe SSD drive would be a member of the device side and the host side refers to the system/OS.

Before removing the drive, you must first notify the host that you would like to remove the drive. This is because the host side is responsible for managing the overall files system and structure. It is also in charge of maintaining data integrity. Once the request has been made, the host side will determine whether the drive is in a safe state. If approved, the host side will then notify the user that it is safe to remove the drive.

The notification step is necessary for ensuring both data integrity and the smooth removal of the SSD. If one were to skip this step, its possible that the drive they removed could have been in the middle of read/write operations, meaning it wasn’t safe to eject. This can lead to serious issues such as data loss or total corruption of the drive.

In order to facilitate the smooth removal of drives and avoid data loss, HighPoint has integrated an unplug feature. This easy-to-use feature can be found using the HighPoint NVMe RAID Management Software, chiefly the WebGUI (Web-Based Graphical User Interface) or the CLI (Command Line Interface). When it comes time to remove the drive, all the user need do is simply use this feature to notify the host of the intent to eject the drive. The user should then see a notification informing them that the drive is now removable.