 |
|
A brief history of database RAM storage
Oracle Database Tips by Donald BurlesonOctober 28, 2015
|
The duties of any database administrator
include responsibility for the management, control and safety of
mission-critical information, and a primary concern of the Oracle DBA is
managing their computer disk storage.
Also see my
brief history of database disk storage.
More than fifty years ago, IBM introduced
RAMAC (random access
accounting & control), the first in a long line of cumbersome and expensive
database storage architectures. Disk platter storage continued to evolve
through the 1990's, when engineers hit the physical limits of mechanical devices
and RAM was introduced as a front-end cache. In the early 21st century we
see RAM-SAN replacing the antiquated mechanical platters.

The introduction of RAM storage
Historically, RAM I/O
bandwidth grows one bit every 18 months, making the first decade of the 21st
Century the era of 64-bit RAM technology:
Decade Bandwidth
1970's
8 bit
1980's
16 bit
1990's
32 bit
2000's
64 bit
2010's
28 bit
As of 2006, the vast majority of hardware vendors (Sun, HP, IBM, UNISYS and
Dell) offer 64-bit servers with far higher bandwidth than their ancient 32-bit
predecessors.
Note that Moore's Law does not apply at all to RAM and the speed has
been fixed for nearly 30 years.
Because CPU speed continues to outpace memory speed, RAM subsystems
must be localized to keep the CPUs running at full capacity.
However, RAM is
quite different from other computer hardware such as disk and CPU.
Unlike CPU speed, which improves every year, RAM speed is
constrained by the physics of silicon technology. Instead of speed
improvements, there is a constant decline in price. CPU speed also
continues to outpace RAM speed and this means that RAM sub-systems
must be localized to keep the CPUs running at full capacity.
?Moore's Law? states that CPU speed will constantly improve while
process costs fall. Unfortunately, this is not the case for RAM and
disk, and we see that the ?real? disk speeds have not improved
significantly in the past 15 years:

Moore's Law, applied to disk speed
CPU speed
continues to double every few years, while the speed of disk and RAM
cannot boast such a rapid rate of speed improvements.
For RAM, the speed has increase from 50 nanoseconds (one billionth
of a second) to two nanoseconds, a 25x improvement over a 30-year
period. At access speeds of two-billionths of a second, today's DDR
SDRAM is stressing the limits of silicon technology, and it's
unlikely that significantly faster speeds will be seen in the next
decade.
Year RAM Type
Access Speed
1987 FPM 50ns
1995 EDO 50ns
1997 SDRAM 15ns
1998 SDRAM 10ns
1999 SDRAM 7.50 ns
2000 DDR SDRAM 3.75 ns
2001 DDR SDRAM 3.00 ns
2002 DDR SDRAM 2.30 ns
2003 DDR SDRAM 2.00 ns
RAM access speed over time
It is very clear than CPU speed will continue to outpace RAM speed and this has
important ramifications for Oracle database processing. The advent of
Non-Uniform Memory Access (NUMA) is predicated on the fact that data storage
(RAM) must be localized as close to the CPU as possible to maximize throughput.
NUMA has been available for years in high-end UNIX servers running SMP
(symmetric multi-processor) configurations. The vendors know that NUMA
technology allows for faster communication between the distributed RAM in a
multi-processor server environment. NUMA is supported by Linux and Windows
Advanced Server 2003 and is a feature of the Intel Itanium2 chipset, which is
used in the latest Oracle server blades for Oracle Grid computing.
In my book "Oracle
Solid State Disk Tuning", smaller solid-state RAM disks have far
less bandwidth issues because the RAM architecture of SSD allow high
concurrent access that is impossible on a mechanical platter.
Oracle 10g has become NUMA -aware and the database engine can now exploit the
high-speed L2 cache on the latest SMP servers. According to David Ensor, a
recognized Oracle tuning expert, author, and Former Vice President of the Oracle
Corporation's Performance Group, the inordinate increase in CPU power has
shifted the bottleneck of many systems to disk I/O, as the disk technology fails
to keep-up with CPU.
For a full discussion, see my whitepaper on
Oracle tuning with solid
state disks.
It's clear that solid-state devices are making huge headway, and
several books on
Oracle
SSD tuning and stepping-up to the new technology. With
prices falling rapidly, I expect that most database will be
solid-state in the next few years and that disk will become the "new
tape", offline tertiary storage for backups.
RAM history References
Market Survey of SSD vendors for
Oracle:
There are many vendors who offer rack-mount solid-state disk that
work with Oracle databases, and the competitive market ensures that
product offerings will continuously improve while prices fall.
SearchStorage notes that SSD is will soon replace platter disks and that
hundreds of SSD vendors may enter the market:
"The number of vendors in this category could rise to several
hundred in the next 3 years as enterprise users become more familiar
with the benefits of this type of storage."
As of January 2015, many of the major hardware vendors (including Sun and
EMC) are replacing slow disks with RAM-based disks, and
Sun announced that all
of their large servers will offer SSD.
Here are the major SSD vendors for Oracle databases
(vendors are listed alphabetically):
2008 rack mount SSD Performance Statistics
SearchStorage has done a comprehensive survey of rack mount SSD
vendors, and lists these SSD rack mount vendors, with this showing the
fastest rack-mount SSD devices:
manufacturer |
model |
technology |
interface |
performance metrics and notes |
IBM |
RamSan-400 |
RAM SSD |
Fibre
Channel
InfiniBand
|
3,000MB/s random
sustained external throughput, 400,000 random IOPS |
Violin Memory |
Violin 1010 |
RAM SSD
|
PCIe
|
1,400MB/s read,
1,00MB/s write with ×4 PCIe, 3 microseconds latency |
Solid Access Technologies |
USSD 200FC |
RAM SSD |
Fibre Channel
SAS
SCSI
|
391MB/s random
sustained read or write per port (full duplex is 719MB/s), with
8 x 4Gbps FC ports aggregated throughput is approx 2,000MB/s,
320,000 IOPS |
Curtis |
HyperXCLR R1000 |
RAM SSD |
Fibre Channel
|
197MB/s sustained
R/W transfer rate, 35,000 IOPS |
Choosing the right SSD for Oracle
When evaluating SSD for Oracle databases you need
to consider performance (throughput and response time), reliability (Mean Time Between failures) and
TCO (total cost of ownership). Most SSD vendors will provide a
test RAM disk array for benchmark testing so that you can choose the
vendor who offers the best price/performance ratio.
Burleson Consulting does not partner with any SSD vendors and we
provide independent advice in this constantly-changing market. BC
was one of the earliest adopters of SSD for Oracle and we have been
deploying SSD on Oracle database since 2005 and we have experienced SSD
experts to help any Oracle shop evaluate whether SSD
is right for your application. BC experts can also help you choose
the SSD that is best for your database. Just
call 800-766-1884 or e-mail.:
for
SSD support details. DRAM SSD
vs. Flash SSD
With all
the talk about the Oracle “flash cache”, it is important to note that there
are two types of SSD, and only DRAM SSD is suitable for Oracle database
storage. The flash type SSD suffers from serious shortcomings, namely
a degradation of access speed over time. At first, Flash SSD is 5
times faster than a platter disk, but after some usage the average read time
becomes far slower than a hard drive. For Oracle, only rack-mounted
DRAM SSD is acceptable for good performance:
|
Avg. Read speed
|
Avg. write speed
|
Platter disk
|
10.0 ms.
|
7.0 ms.
|
DRAM SSD
|
0.4 ms.
|
0.4 ms.
|
Flash SSD
|
1.7 ms.
|
94.5 ms.
|
|