A thorough explanation of the basic technology of SSDs: the basics of their mechanisms, structure, and selection points

How SSDs work

An SSD is a storage device that primarily consists of NAND FLASH MEMORIES, DYNAMIC RAMS cache, and an advanced controller.
NAND flash stores data electrically, eliminating the need for physical moving parts, enabling fast data Access and high reliability. Its simple structure also makes it highly shock-resistant. SSD controllers not only optimize data read and write, but also incorporate advanced management functions such as error correction (ECC), wear leveling, and garbage collection.

SSD configuration

Advantages of SSDs

The benefits of introducing SSDs are wide-ranging. In data centers, they shorten server boot times and improve the throughput of AI inference and big data analysis. In industrial equipment, they improve the response of real-time control and data collection, speed up recovery from failures, and improve the work environment through quieter operation. In the automotive field, they contribute to shorter ECU boot times, faster software updates via OTA (Over The Air), and improved real-time processing performance of ADAS and autonomous driving systems.

Another major benefit is power efficiency. SSDs consume less power than HDDs, helping to reduce power costs in data centers, extend battery life in industrial and automotive equipment, and reduce heat generation. Shock resistance and high reliability are also important factors that support stable operation in harsh environments.

M.2 SSD

Choose the best interface and form factor for your application

To maximize SSD performance, it is necessary to select the interface and form factor that best suits your application.

What is an SSD interface?

An SSD's interface refers to the data communication standard and physical connection method that connects the storage device to the system. There are two types of SSD interfaces: PCIe and SATA. The two interfaces are not compatible with each other.

SATA (Serial ATA)

SATA is an interface that has been widely used in conventional HDDs and SSDs. SATA 3.0 has a maximum transfer speed of approximately 600MB/s. With few signal lines and a simple design, it is suitable for applications where cost and compatibility with existing design assets are important.

SATA

PCIe (NVMe)

PCIe (Peripheral Component Interconnect Express) is a high-speed interface that is mainstream in PCs, servers, the latest industrial equipment, and automotive computers. NVMe (Non-Volatile Memory Express) is a protocol for controlling SSDs over PCIe, and with PCIe Gen6, a maximum transfer speed of approximately 30.25 GB/s is achieved using x4 lanes. PCIe bandwidth varies depending on the number of lanes and generation.

PCIe M.2

What is an SSD form factor?

The form factor refers to the physical shape, size, and CONNECTORS type of an SSD. In design, it is an important factor that directly affects installation space, cooling efficiency, maintainability, and other factors.

Main form factors

• 2.5-inch SATA: Supports SATA connections and is widely used as an alternative to HDDs.
• M.2: Compact and thin, it's ideal for space-constrained applications such as laptops, embedded devices, and in-vehicle computers. NVMe-compatible models use PCIe slots for high-speed data transfer. SATA-compatible models are also available.
• U.2: A standard for 2.5-inch SSDs, it is used in enterprise applications and is suitable for servers and industrial storage that require high durability and high performance. It also features hot-swap compatibility and easy maintenance. As the name "universal" suggests, the U.2 form factor is compatible with various interfaces such as PCIe and SATA.
• E1.S / E3.S: Form factors for next-generation data centers that enable high-density mounting and highly efficient cooling designs. Applications are also expanding to highly integrated boards for industrial equipment and automobiles.

Various form factors

How to choose a NAND type: balancing durability and cost

NAND flash, the storage medium for SSDs, comes in several types, including SLC, MLC, TLC, and QLC. The diagram below shows the types of data that can be stored per cell. The simpler the structure, the higher the reliability but the higher the cost, while the more complex the structure, the lower the reliability but the lower the cost.

NAND flash type (SLC, MLC, TLC, QLC)

• SLC (Single Level Cell): This type records one bit per cell and is characterized by extremely high rewrite durability (100K P/E cycles). It is suitable for industrial equipment, mission-critical applications, and data center log storage, but the cost per unit capacity is high, so increasing the capacity is costly.
• MLC (Multi-Level Cell): Records two bits per cell and offers better cost performance than SLC. It also has high durability (10K P/E cycles), making it widely used in general industrial equipment, automotive applications, and data center storage.
• TLC (Triple Level Cell): Records three bits per cell, offering a good balance between cost and capacity, and is the mainstream in client PCs and many enterprise SSDs.
• QLC (Quad Level Cell): Records 4 bits per cell, and is characterized by its large capacity and low cost. It is mainly used for read-based applications and cost-conscious storage, but because its rewrite durability is lower than other methods (1K P/E cycle), it is important to select a cell that suits the actual operation. It is suitable for data center archives, long-term storage of industrial equipment, and in-vehicle map data storage.

When selecting an SSD, it is important to understand the workload (write volume, random Access ratio, number of simultaneous Access at peak times) and comprehensively evaluate the TBW (total rewritable capacity), warranty period, and controller wear leveling performance.

The role of the SSD controller

The controller, which can be said to be the heart of an SSD, is an important component that greatly affects the overall performance and reliability of the SSD. The controller optimally controls the writing and reading of data to the NAND FLASH MEMORIES, thereby bringing out the inherent high speed and long life of the SSD.
The controller is responsible for data management, controlling the efficient placement and Access of data in NAND cells. This significantly improves read/write speed and response. The controller also handles ECC (Error Correction Code), which automatically detects and corrects errors that occur within the SSD, improving data integrity and reliability.
Furthermore, because NAND flash has a limited number of rewrite cycles, the controller uses a technology called Wear Leveling to distribute writes evenly across the entire SSD to prevent certain cells from being overused and prematurely deteriorating, thereby extending the lifespan of the SSD as a whole.
Another important controller function is support for the TRIM command. TRIM allows the OS to notify the SSD of unused areas due to deletion or other reasons, streamlining the organization of unnecessary data (garbage collection). This reduces overhead when writing, preventing degradation of SSD performance and an increase in the number of writes.
As such, the controller plays a central role in supporting the performance and reliability of the SSD, and the performance and functionality of the controller are important points to consider when selecting an SSD.

SSD technology that enhances reliability and security

Reliability and security are essential elements of any modern storage system, and SSDs incorporate a variety of technologies to ensure data safety and longevity.

• Power loss protection (PLP)
  PLP is equipped with dedicated circuits and CAPACITORS to protect data even in the event of a sudden POWER SUPPLIES outage and reduce the risk of losing data being written. This ensures that important information being stored is not lost even in the event of a system failure or unexpected power outage. This is particularly effective in systems that record data in real time or in applications that require long-term continuous operation.
• SMART/Telemetry
  SMART (Self-Monitoring, Analysis and Reporting Technology) and telemetry functions constantly monitor and record the health status of storage, lifespan prediction, error occurrence status, etc. This allows for early detection of signs of failure and enables planned and preventative maintenance.
• Hardware encryption (AES-256)
  The AES-256 hardware encryption function is a mechanism that automatically encrypts data when it is written inside the SSD. Because encryption is performed at the hardware level, it reduces the load on the host CPU. Even if the storage is removed from the outside, the data cannot be decrypted without the encryption key, which significantly reduces the risk of unauthorized Access or information leaks. In particular, for systems that handle highly confidential data or personal information, the presence or absence of hardware encryption is a major factor in deciding whether to implement it.
• SED (Self-Encrypting Drive)
  SEDs (Self-Encrypting Drives) are SSDs that automatically encrypt and decrypt data. SEDs comply with industry standards such as TCG Opal, and can meet the strict security requirements of companies and organizations. A distinctive feature of SEDs is that data stored in them cannot be Access without a password or authentication information set by the user or administrator. They also have a cryptographic erasure function, which allows stored data to be quickly and securely erased when the drive is reused or disposed of. This minimizes the risk of information leaks, allowing for worry-free operation.
• Firmware Robustness
  Firmware is the heart of storage, controlling its operation and security functions, and regular updates and vulnerability countermeasures are essential. SSDs equipped with signed firmware and secure update functions prevent unauthorized Access and tampering from outside, allowing for safe use even over the long term. Recently, remote updates via OTA (Over The Air) have become common, making it increasingly important to balance security and reliability in the field.

As such, SSDs use multiple layers of technology to enhance reliability and security. When introducing SSDs, it is important to comprehensively evaluate the required features, such as PLP, SMART, encryption and erasure functions, and firmware robustness, depending on the operating environment and application.

SSD Technology Trends: Multi-tiering, PCIe Gen6, and CXL Integration

SSD technology is evolving rapidly, with the increasing number of layers in NAND FLASH MEMORIES accelerating significantly. In recent years, ultra-multilayer NAND structures have emerged, ranging from 232 layers to 276 layers and even over 300 layers, dramatically increasing SSD storage capacity while also significantly improving data read/write performance. The increased density achieved by multilayering not only allows more data to be stored in a limited space, but also brings significant benefits in terms of Access speed and power consumption.

PCIe Gen6 SSDs, defined by the PCI-SIG (PCI Special Interest Group), achieve bandwidth exceeding 128 Gbps, significantly exceeding previous generations, and demonstrate their true value in cutting-edge fields that require high-speed, simultaneous processing of massive amounts of data, such as AI and High Performance Computing (HPC). In applications such as AI model training and inference, and big data analysis, data read/write speeds often become a bottleneck, and the introduction of PCIe Gen6 SSDs dramatically improves the processing efficiency and responsiveness of the entire system.

SSDs and CXL (Compute Express Link) are a new combination that is gaining attention in the fields of next-generation data centers, AI, and HPC. CXL is a high-speed interface based on PCe, allowing for more flexible connections between CPUs and accelerators than conventional NVMe SSDs. This allows SSDs to be treated not just as storage but like system memory, allowing the CPU to directly and quickly Access data on the SSD. As a result, I/O bottlenecks are significantly alleviated, enabling more efficient system designs for applications requiring large volumes of data and high-speed processing, such as AI and big data analysis. Various manufacturers are developing CXL-compatible SSDs as a key technology that will support the future evolution of data center infrastructure.

Summary

Understanding the basic technology of SSDs will enable you to select the appropriate storage for your application and operating environment. Understanding the basics, such as the types of NAND flash, the role of controllers, and the characteristics of interfaces and form factors, will lead to stable system operation and improved performance. First, you need to fully understand the mechanism and structure of SSDs before selecting one that meets your company's needs.