In this guest blog post, Flash Drives: How to Make the Right Choice, Innodisk provides an overview of the architecture and structure of flash memory and advises on things to consider when a selecting an industrial flash-based SSD.
Contents
- NOR and NAND architectures
- 3D NAND, the second generation of NAND flash
- Memory cells
- Choosing a flash drive using Innodisk solutions as an example
- Form factors
- Operating conditions
- Resource and speed
- How to read Innodisk model names
- The Innodisk SSD series
NOR and NAND architectures
There are two flash memory architectures – NOR and NAND – differing primarily in the method of connecting memory cells. Each of them has its own advantages and disadvantages, which governs how and where they are best used.
NOR is from the logic expression ‘Not OR’, and the NOR memory architecture uses a classic two-dimensional conductor matrix with one cell at the intersection of rows and columns. With this arrangement, NOR provides high-speed, random access to information and can write and read data at a specific location without having to access memory sequentially. Unlike NAND memory, NOR memory can access data up to one byte.
NOR technology is an advantage in situations where data is being written or read randomly.
NOR is therefore more often used as program memory for microprocessors (in the BIOS memory of a personal computer, for example) for storing a small volume of auxiliary data, or for storing an operating system in a cell phone or tablet.
NAND memory on the other hand is named after logic expression ‘Not AND’. Basically, the matrix is the same as in NOR, but instead of one cell (transistor), a column of cells in series is found at each intersection.
NAND memory writes and reads data at high speed, in sequential read mode, arranging data into small page blocks.
It can read and write information page-by-page but, unlike NOR, it cannot access a specific byte.
The price of an SSD using NAND architecture is less than NOR. Also, NAND chips have a higher density of data. NAND architecture is commonly used in SSDs, USB drives, memory cards, mobile phones for storing user information, and other devices in which data is written sequentially.
In this blog post, we will mostly focus on the NAND architecture as it is used in most modern solid-state drives.
3D NAND
This was devised by manufacturers because the 2D approach (even though fabricated at geometries as small as 10nm) could not support their roadmaps for increasing memory density. Adding a third dimension, height, made sense.
Memory Cells
A memory cell can store one, two, three or four bits of information. Physically, all four types of NAND memory cells are composed of the same transistors, only the amount of charge stored by the cell differs.
In single-bit cells there are only two charge levels on the floating gate. Such cells are called single-level cells (SLC). SLC memory has the highest price, highest performance, lowest power consumption, highest write speed and program/erase (P/E) cycles.
More charge levels are found in multi-bit cells; that’s why they are called multi-level cells (MLC). MLC devices are cheaper and have more capacity than SLC devices, but they also have a higher access time and their maximum number of rewrites is much lower. Note: MLC is typically used to denote four charge levels (i.e., two bits) per cell. This is also sometimes called double-level cell (DLC).
Memory with eight charge levels (i.e., 3-bit) is called TLC (Triple Level Cell). It lags further behind MLC (and certainly SLC) in terms of read/write speed and the number of P/E cycles. Only plus side, density is higher.
Not shown in the above diagram, but quad- and penta-level cell (QLC and PLC, respectively) is also available. QLC has 16 charge levels and can store 4 bits per cell, and PLC has 32 charge levels and can store 5 bits per cell. Again, read/write speed and the number of P/E cycles suffers as a consequence of the greater storage capacity (and low cost).
Tiered memory dominates the market today. Nevertheless, SLC-products, despite the many times lower capacity and high cost, continue to be developed and manufactured for critical applications, because of the fast read/write speed and high number of P/E cycles.
Note: Innodisk recommends only SLC or technologies based on MLC (such as iSLC) or TLC (such as 3D TLC WD NAND) for use in applications where reliability is key. Indeed, the company has no plans to make QLC or PLC NAND.
Choosing a flash drive with Innodisk solutions as an example
The choice of solid-state drive for your computer depends on several factors, the most important of which are: form factor, scope of application, operating conditions, speed and resources.
Form Factors
The process of selection of any storage device begins with the slot and connector to install it. There are many form factors and interfaces available on the market today for connecting flash drives. It’s safe to say that the most common is the 2.5-inch form factor with a third-generation SATA connection interface.
A 2.5-inch SATA connector can be found in almost every modern personal computer or laptop and is universal for installing a classic hard drive, as well as a solid-state drive based on flash memory.
Available form factors:
Industrial CompactFlash cards
CompactFlash memory cards with high read/write speed and capacities from 256MB to 64GB. One of the oldest types of flash memory. |
|
Industrial CFast cards
Industrial CFast memory cards with a standard 7-pin SATA interface and a 17-pin power connector (different than SATA one). |
|
Industrial SD cards
Secure Digital memory cards are designed for use in portable devices, mini PCs and DIN-rail mounted computers. Equipped with a mechanical write protection switch. |
|
mSATA SSD
Modular drives designed for installation in Mini-PCIe expansion slots: full-sized mSATA modules and half-sized modules. |
|
SLIM SATA SSD
Compact electronic drives with standard SATA interface (7 + 15-pin). Due to its smaller size, SLIM SATA SSD is an excellent replacement for 2.5-inch drives when mounting space is limited. |
|
1.8″ SATA SSD
Solid state drives supporting micro SATA (uSATA) interface. The micro SATA connector consists of two independent contact groups: 7-pin for data bus connection and 9-pin for power connection. |
|
2.5″ IDE SSD
Available with different interfaces: PATA and IDE. If you need to update your legacy system, then installing an SSD with IDE interface instead of a standard HDD will increase the system performance significantly. |
|
2.5″ SATA SSD
Solid State Drives with SATA 2 and SATA 3 support. Used in compact devices and in industrial systems to improve performance. |
|
M.2 SSD
Compact and versatile solid-state expansion cards exhibiting high write speeds, containing SATA and PCI-E interface components. Designed for a diverse set of tasks. |
|
Mini-PCIeDOM
Flash based storage media with Mini PCI Express interface. |
|
Disk on Module
Horizontal, vertical, with or without housing, low profile and rotatable DoM modules designed to be installed directly on the mainboard or with SATA/IDE cables (40 and 44-pins). |
|
OCuLinkDOM
Ultra-compact solid-state drive for server solutions with OCuLink connection interface, operating by NVMe protocol via PCI Express Gen3 bus. |
Operating Conditions
An industrial computer differs from its customer-class counterpart in many ways, but the most prominent differences are the capability to operate in 24/7/365 mode without interruptions or downtime, resistance to temperature spikes, vibrations, shock and high humidity. These demanding requirements to an industrial computer apply to all of its components, including storage devices.
Depending on the degree of protection against these factors, the price of a flash memory device also differs. Therefore, it is important to determine in what operating conditions the final product will be used and what temperatures and spikes it will be exposed to.
Flash memory drives can be divided into two major classes:
-
- Standard temperature grade models
- Extended temperature grade models
Resource and Speed
As we have already said, there are flash memory cells capable of storing between 1- and 5-bits of data (SLC to PLC). In industrial Innodisk flash memory, only SLC, MLC/DLC and TLC are used – due to stringent requirements for reliability.
The algorithm for processing a charge consisting of 16 levels (i.e., 4-bit) and 32 levels (5-bit) is too complex, and in conditions of voltage surges and possible radio frequency interference that may occur during the operation of an industrial computer, this can lead to a rapid decrease in the residual resource and, as a result, data loss.
In addition to the three main types of memory cells, there are enterprise MLC cells – eMLC. Innodisk calls these cells iSLC; they are the middle ground between the endurance of SLC and the more affordable price of MLC.
iSLC drives demonstrate performance and reliability comparable to SLC products, but are built on MLC memory hardware, making them a more affordable solution suitable for servers and enterprise segment systems.
When using SSD based on iSLC technology, you can be sure of data safety – the service life is increased several times compared to standard MLC products.
Here is a comparison table of memory cells used in flash drives, in terms of price/endurance/speed. Let’s compare four models of Innodisk solid-state drives of 2.5-inch SATA form factor with a capacity of 128 Gigabytes, an extended temperature range of -40 to 85oC, and designed to be used in an embedded system.
How to read Innodisk model names
Large industrial flash memory vendors such as Innodisk have a wide variety of devices with similar specifications across a range of product lines.
In order to find your way across this abundant selection and choose a drive most suitable for your specific tasks, you first need to figure out what kind of SSDs are available and what they are geared towards.
As an example, we will look at the model name of a 2.5-inch SATA SSD:
(2.5″ SATA SSD) – form factor: points to the flash storage device design.
(3MV2-P) – series name. Let’s take a closer look each symbol in turn and what it represents:
(3) – generation: 1, 2, and 3.
First, second and third generation models are available.
(M) indicates the NAND type: S, M, I, or T.
(V) – scope of application, with the following series available: E, G, R, V, S.
(2) – production series.
This number can range from 1 to 10, or there may be no at all. It has no significance when choosing an SSD-drive.
(-P) indicates the presence of a DRAM-based cache.
The clipboard (cache) is used primarily for storing the address translation table, which increases the speed of accessing flash memory and writing files.
For every 1GB of SSD storage, there must be 1MB of cache. Thus, an SSD with a capacity of 120-128GB must have 128MB cache, a 240-256GB SSD must have 256MB cache, a 500-512GB SSD must have 512MB cache and a 960-1024GB SSD must have 1024MB cache.
SSD without a DRAM buffer will slow down during long write operations of small files.
Innodisk SSD Series
In addition to the series listed above, Innodisk offers a unique product – the fireproof 3.5-inch Fire Shield SSD. It is capable of withstanding a temperature of 800oC for 30 minutes, protecting the memory elements from external influences.
Conclusion
In this blog post, we have:
- Learned what an SSD is.
- Given a definition of flash memory and explored what NOR and NAND architectures are.
- Figured out why the NAND architecture is better suited for SSD drives.
- Found differences between the types of NAND memory cells (SLC, MLC, TLC, QLC and PLC).
- Compared several Innodisk models, finding out which types of cells have the greatest durability and reliability, and which ones are affordable.
- Defined what form factor is and how they are used in modern industrial PCs.
- Analysed which series of products exist and for which tasks they are best suited, using Innodisk SSD series as an example.
We hope this material is useful in easing your way into the vast world of industrial flash memory drives.