DRAM vs. DRAM-less SSDs: Key Differences and How to Choose the Right One

The Solid-State Drive (SSD) has become incredibly popular as the world of data processing and storage changes with new IoT and 5G applications and their design performance requirements.

The Serial Advanced Technology Attachment (SATA) interface protocol, which was used by legacy SSDs in the early 2000s, restricted the bandwidth and speed at which storage devices could transfer data to a motherboard. Non-Volatile Memory Express (NVMe), which employs the Peripheral Component Interconnect Express (PCIe) interface and enables speeds that far above SATA norms, is now widely used by manufacturers. For sophisticated industrial-grade SSDs, NVMe is swiftly taking the lead as the storage interface.

In addition to Dynamic Random Access Memory (DRAM), which is around 100 times faster than NAND flash memory, SSDs use NAND Flash memory, the most costly but essential part of the device, as a cache for writing data to the SSD and for storing and recording its data mapping tables. SSD makers consider design options, such as DRAM-less SSDs, to increase performance, lower prices, and lessen the negative effects of DRAM supply/demand limits.

DRAM vs. DRAM-less SSDs: Definition

Data mapping tables, which maintain track of logical blocks and their physical positions in NAND, are stored in the DRAM of an SSD. On SSDs without DRAM, this mapping table is kept on NAND.

DRAM-less SSDs don’t have an inbuilt DRAM chip, as the name suggests. Rather, they employ different techniques, like making use of the RAM on the host machine or using on-board NAND flash memory for caching.The performance of SSDs without a DRAM cache may be slower than that of SSDs with a DRAM cache because NAND is slower.

DRAM vs. DRAM-less SSDs: Purpose

The main purpose of DRAM is to store the FTL when the SSD is turned on. The FTL is kept on the permanent NAND flash memory while the system is turned off. Because DRAM has a shorter latency and a quicker access time than NAND storage, it is present. What, then, is FTL? It is an address mapping table from logical to physical. FTL is responsible for translating the operating systems’ requests into the actual NAND flash locations where the data is kept. Now, the FTL’s speed has a significant effect on performance for both writing and reading. For this reason, by acting as a speedier storage media than the actual NAND flash, DRAM helps your SSD operate better.

DRAM vs. DRAM-less SSDs: Benefits and Drawbacks

Benefits

Write buffering is significantly influenced by SSD DRAM.  The technique known as “write buffering” involves temporarily storing newly introduced data in a buffer before it is physically written to its storage destination.  Due to SSD “erase-before-write” limitations, DRAM serves as a write cache and uses its high read/write performance to temporarily manage incoming data bursts.

Because DRAM and RAM are volatile, information is lost when the power is cut off. SSDs are fantastic because of their “journal commits” and capacitors, which guard against data loss in the event of a power outage. In these cases, the data remaining in the DRAM is transferred to the NAND flash for long-term storage in the event that the system loses power. That data is reloaded into the DRAM when the system boots up again.

Additionally, by enabling quicker map access, DRAM aids with garbage collection, heap-leveling, and data prefetching, which increases read speed. When the SSD is idle, all of the algorithms typically operate in the background. As a result, these background activities have little to no effect on SSD performance when DRAM access is available.

Drawbacks

It uses more system resources: Although DRAM is renowned for its speed, maintaining it requires a significant amount of power. It’s a hotter kind of memory since DRAM generates greater heat due to its power consumption, which could damage other computer components.

The main drawbacks are increased write amplification, slower performance in random read/write activities, and higher latency. Additionally, DRAM-less SSDs perform less reliably over extended periods of time.

DRAM vs. DRAM-less SSDs: How they work

An SSD transfers data onto those memory cells when it saves it, but each time data is stored to a cell, the cell deteriorates slightly. The SSD uses a technique known as wear levelling to transfer data to various cells in order to prevent one cell from deteriorating more quickly than the others. As data continuously shifts from cell to cell, the DRAM within the SSD keeps track of where each bit is stored. When data is required, the SSD controller quickly calls up and makes it available by looking at the mapping table on the DRAM.

The majority of SSDs have DRAM, although an SSD without DRAM does not necessarily lack a data map. The map data on a DRAM-less SSD is typically stored in the host memory buffer (HMB) of the device utilising the SSD (e.g., an enterprise server in a data centre or a tablet used by a mobile worker) or in the SSD’s NAND flash memory itself (using the tiny bit of SRAM space on the controller).

DRAM SSDs vs. DRAM-less SSDs: Use Cases

Applications like gaming, content creation, and database administration that need fast random access are ideal for DRAM SSDs. They are particularly good at jobs that need frequent and varied data access patterns.

Lacking DRAM SSDs are frequently used when performance requirements are lower and affordability is a major factor. While their performance might not be as reliable in some demanding workloads, they can be appropriate for routine computing activities.

In conclusion, the decision between DRAM SSDs and DRAM-less SSDs depends on the user’s unique needs and financial constraints. While DRAM-less SSDs offer a more affordable option for less demanding applications, DRAM SSDs are generally preferred for high-performance computing tasks.

DRAM vs. DRAM-less SSDs: How to choose

For consumers who do not require high-sustained performance over an extended period of operation, DRAM-less SSDs offer a compelling substitute. Additionally, DRAM-less SSDs offer a cost-effective alternative and are immune to seasonal shortages of DRAM, which can unpredictably raise component costs and the availability of DRAM-enabled SSDs.

Today’s NVMe SSDs’ lack of integrated DRAM also allows for improved thermal management, reduced power consumption, and effective PCB layout.

The Host Memory Buffer (HMB) was activated for users who wish to combine better performance with a cheaper SSD cost structure. Through the PCIe connection, HMB deployment enables DRAM-less SSDs to use a portion of the DRAM connected to the system’s Central Processing Unit (CPU), thereby reducing costs and improving performance metrics.

FAQ