CompactFlash is a decades-old format with a long industrial history, but the availability of modern controller ICs means that CF doesn’t have to use old Flash technology

By Lena
Harman, Hyperstone GmbH

While the consumer market pursues
the latest and greatest high-performance SSDs, industrial markets, where
reliability trumps performance, are still using the CompactFlash form factor that
was introduced decades ago. But decades-old form factors don’t necessarily have
to use obsolete technology. Advances in Flash Translation Layer (FTL) and
mapping algorithms, wear leveling, and error correction coding (ECC) have made
it into the latest controllers designed for CompactFlash cards, where they allow
these old form factors to use the recent, more cost-efficient Flash devices and
provide larger capacities.

Enduring legacy
First introduced in 1994,
CompactFlash has gone through several iterations and improvements that have
kept it relevant even today, 23 years later. Compatible with PCMCIA and IDE
interfaces, Compact Flash has been used since the 1990s as a simple way to
integrate Flash memory into a mass storage system. While today’s client and
enterprise storage devices mainly use SATA- and PCIe-based interfaces rather
than PCMCIA and IDE, CompactFlash and PATA still have a big presence in
industrial markets. CompactFlash is also widely used in top digital single-lens
reflex (DSLR) cameras from Canon and Nikon, as well as networking routers and
industrial equipment.

What’s the secret behind the
enduring popularity of CompactFlash? Simply put, CompactFlash provides an easy-to-use,
removable, durable Flash drive. With its robust case, it is heftier but also
more rugged than the thin and flimsy SD cards with which it competes. In
addition, unlike traditional 2.5-inch PATA, SATA, or today’s M.2 PCIe Flash
drives, CompactFlash is designed to be installed and removed repeatedly.
Without any training, an operator can switch out the Flash drive on a router or
a piece of industrial equipment to upgrade its firmware or copy its contents.
And unlike traditional hard drive form factors, there’s no separate power cable
to fumble with because power is provided through the pins.

Hyperstone-CompactFlash-drive

The robust
body of CompactFlash drives make them more suitable to rough industrial
applications than the SD card format. (Source: SanDisk)

CompactFlash
today
The uses of CompactFlash can be
divided into legacy and modern applications. In legacy systems, CompactFlash
drives are being used either as first-generation Flash storage devices or to
replace hard drives. These older systems often don’t require (and can’t take
advantage of) faster speeds; they simply require a stable storage solution with
a reliable source of replacements.

For these systems, product
longevity is more important than raw performance. While consumer product
lifecycles are often measured in months, industrial storage needs to remain in
use and available from vendors for years, even decades. For these systems,
then, storage solutions need to maintain consistent interface and operational
characteristics for long periods. The key component for providing this
interface is the Flash controller, so devices like Hyperstone’s F2 Flash
controllers, which have been produced for 15 years and counting, make the most
sense for designers.

Even though the interface and
operational characteristics need long-term consistency, however, CompactFlash
drives don’t have to be stuck in time with all the limitations of earlier Flash
device designs. Modern CompactFlash drives can offer performance, capacity,
security, and reliability that rivals today’s SATA drives while still providing
the ruggedness as well as ease of installation and removal of traditional CompactFlash.
This is possible because advanced CF controllers like Hyperstone’s F9 series
offer designers many features to help leverage modern Flash technologies.

Hyperstone-F9-CF-Controller

Advanced CF
controllers like the Hyperstone F9 series support the use of modern Flash
memory devices in industrial applications.

Enhanced wear leveling
One of these key features is
enhanced wear leveling. Service life is an essential factor for industrial Flash
drives, and because Flash memory devices have a limited number of write cycles,
improving endurance can significantly reduce maintenance costs and lengthen
operation lifetime for industrial equipment. The main mechanism that CF
controllers use to lengthen service life is wear leveling. By spreading wear
around the drive instead of repeatedly writing to the same locations, the
maximum lifetime of an SSD can be achieved.

Page-based mapping FTL
A second key new feature involves
the Flash Translation Layer (FTL), which maps the logical storage (file names)
to physical device locations. All Flash storage systems need to map file system
accesses to physical Flash accesses, but most removable Flash drives,
especially older ones, map data in a so-called block-based mapping approach. This
relatively large-grain mapping is easier to implement than other approaches,
and in the past, Flash block sizes were relatively small, so the block-based
mapping overhead was acceptable.

However, block-based mapping is
far from ideal today because Flash device capacity has increased dramatically,
resulting in larger block sizes that give rise to an endurance problem. While complex
in detail, put simply, block-based mapping results in frequent block erases due
to remapping activity, even if only a small amount of data has actually been
written by the system. This is referred to as “high write amplification,” and
the result is greater device wear.

Newer Flash controller technology,
like hyMap
from Hyperstone, performs data access at a much finer-grain level, smaller than
a block or even a page. This is a more sophisticated method that requires more
processing by the Flash controller, but it results in significantly better
service life for the drive. By mapping (and thus manipulating) data at this
more granular level, fast drive wear-out is avoided and endurance is greatly
improved.

The page-based mapping FTL, along
with modern BCH error correction that is compatible with recent Flash
technology, allows for higher-capacity and lower-cost MLC or 3D NAND to be used
in CompactFlash drives, even in industrial applications. This ability to use
newer, higher-density Flash, in turn, not only improves capacity but also
allows for higher-throughput devices. One drawback, however, is that Flash
program times have increased, slowing things down, especially for small data-write
operations. But again, advances in controller and interface technology as well
as increased buffer space have compensated, enabling industrial-grade
CompactFlash cards to achieve 120-MB/s read/write speeds.

DRAM-less operation
Another feature of modern CF
controllers relates to data buffering. SSDs often use a temporary buffer for
their mapping data and firmware tables, noting, for instance, which blocks of
the Flash have been erased and how many times. Most SSDs temporarily store this
information in volatile memory such as DRAM. This presents a significant reliability
risk, however; a power failure at the wrong time could erase this important
metadata.

Industrial SSDs using DRAM
buffers thus need robust power protection. Some SSDs use supercapacitors as a
backup power source, but this is an expensive approach, and capacitors can fail
over time. A better approach is to avoid DRAM and its need for backup power by
storing mapping information entirely in non-volatile memory. By storing mapping
information away from volatile DRAM, drives can prevent serious data loss due
to power failure. Combined with a transaction-based log, drives can ensure that
data is always written to Flash successfully and maintain mapping table
coherence.

Aside from the risk of failure
and decreased robustness, using DRAM also poses a significant sourcing risk.
The fast product lifecycles and volatile supply that DRAM has historically
suffered can contribute to faster obsolescence and higher price volatility in
SSDs as well.

Health monitoring
Modern CF cards also allow for
health monitoring of the Flash drive using self-monitoring, analysis, and
recording technology (SMART). Operating systems that support SMART can check
the spare block count, block erase counts, total number of logical block
addresses (LBAs) written, and other important drive health information. This
allows users to monitor the service life of their drive as it is being used and
replace it well before the end of its life.

Industrial endurance
While consumer and enterprise
applications move onto other form factors, then, CompactFlash still lives on in
many legacy and modern industrial applications. From telecom towers to
networking routers to industrial equipment, the rugged and easy-to-install
nature of CompactFlash has made it incredibly enduring. Today, CompactFlash
drives are available that can fulfill legacy requirements while utilizing
state-of-the-art technology. The latest CompactFlash controllers enable creation
of fast, high-capacity drives with better performance and reliability features
than ever. Sometimes old dogs can learn new tricks.

Hyperstone-Lena

Lena
Harman is responsible for digital marketing, online strategy and the
optimization of online platforms at Hyperstone. She holds a double degree in
Communications and International Studies from the University of Technology,
Sydney.