IaC & GitOps in a Nutshell - a FridayInANuthshell Episode.pdf
Advantech Platforms For 10 Gig E Networking On Xeon 5500 V1
1. Xx
Advantech Platforms for 10 GigE networking
on Intel® Xeon® Processor 5500 Series
While we have numerous methods of measuring performance in terms of absolute values it is also a
matter of perspective and relativity. Driving at 70mph on the freeway is very fast relative to our own
walking or even running pace but pales in comparison to a 200mph race car, let alone a Mach2 fighter
plane. A chain is only as strong as its weakest link and as such we are always striving to improve that
“weak link.” So it is with any computer or network system platform. When we had to deal with a mere
10 or even 100Mbps on the network side, processors had no issue keeping up. As we moved to GigE
with first 1 and now 10 GigE the weak link or bottleneck would switch between processing capacity and
network bandwidth. With the introduction of Intel's latest‐generation microarchitecture (Nehalem) and
the Intel® Xeon® Processor 5500 Series we are witnessing the next step in the evolution of parallel multi‐
core processing, providing unprecedented performance and dynamic scalability. Enabling the next
generation of packaged application server platforms and ATCA blades, Nehalem processors will be there
to “suck up” that 10 GigE traffic. The perfect solution for performance hungry, high bandwidth, deep
packet inspection, security and other content aware applications.
Processor Evolution
When it comes to examining the evolution of processor performance or for that matter most of the
technology that now surrounds us Moore’s law comes to mind. Gordon Moore, a co‐founder of Intel,
observed back in 1965 that the number of transistors that could be crammed into an integrated circuit
was doubling approximately every two years. We have made great leaps in our processor innovation
and no longer simply add more and more resistors; however, Moore’s law has been expanded and is
often used to describe the overall evolution of performance, size, capacity and price. For example a PC
will halve in price for a technology that may be 2 years old and the price for latest models remains static
while performance and capacities double. The Intel® Xeon® Processor 5500 Series is the fastest yet.
How Fast is Fast Enough
Why faster and faster, won’t there be a point in time when we have all the speed we need? Over the
years numerous industry pundits and economists have claimed, on more than one occasion, that
Moore’s law and the “need for speed” were due to become irrelevant. The claims that computing and
new distributed architectures had become as fast as necessary making further processor enhancements
moot reminds us of the often quoted Charles H. Duell (Commissioner, U.S. patent office, 1899), who was
reputed to have claimed, “Everything that can be invented has been invented.” Before we move on to
look at Intel’s Nehalem (Intel® Xeon® Processor 5500 Series) and why there are still numerous
applications that will benefit from increased speed and capacity its worth considering one last law –
Parkinson’s law. Originally coined in an essay published in The Economist in 1955, Cyril Parkinson stated
that “Work expands so as to fill the time available for its completion.” Having taken on numerous guises
there are variations clearly appropriate for the computing & communications industry. “Data expands to
fill the space available for storage,” or “Network traffic will fill available bandwidth” and of course
“applications will grow to exceed the capabilities of processor performance.”
www.advantech.com/NCG
2. Xx
We can all think of examples, both personally and professionally, where computer systems we have
owned that previously performed just fine, ground to a halt when new and “upgraded” versions of
application and operating system software were installed. That just shows Parkinson’s law at work,
luckily we have Moore’s law that comes to the rescue! There will always be a new application that
requires extra horse power to work at its best and our need to enable greater functionality in smaller
footprints has no end in sight. Hence the welcome introduction of the next evolutionary step in
processor technology. With up to four high performance cores, greater parallelism for increased
instructions per cycle, simultaneous multi‐threading, potentially doubling the number of executing
threads per processor, and built in application accelerators, Intel’s Tylersburg‐EP/Nehalem‐EP Platform
provides an abundance of speed, power and bandwidth for the most demanding of applications.
TylersburgEP/NehalemEP Platform
The Intel® Core™ i7 processor, the first example of Intel’s new microarchitecture (Nehalem), hit the
streets last year with screaming performance. Based on the same Intel® 45nm hi‐k metal gate silicon
technology, there is now a range of extended life processors ‐ the Nehalem EP. With 4 primary versions
including two with low power envelopes, the Tylersburg‐EP/Nehalem‐EP Platform with its parallel
processing engine, represents the perfect marriage of superb performance, scalability and energy
efficiency. The Intel® Microarchitecture (Nehalem) includes:
Dynamic scalability, managed cores, threads, cache, interfaces, and power for energy‐efficient
performance on demand.
Intel® Turbo Boost Technology delivers additional
performance automatically when needed by taking
advantage of the processor's power and thermal
headroom. This enables increased performance of
both multi‐threaded and single‐threaded workloads.
Intel® Hyper‐Threading Technology brings high‐
performance applications into mainstream computing
with 1‐16+ threads optimized for a new generation
multi‐core processor architecture.
Figure 1. Intel®Turbo Boost Technology
Design and performance scalability for complex,
performance hungry application with support for
dual and quad cores and up to 16+ threads with
Intel® Hyper‐Threading Technology (Intel® HT
Technology), and scalable cache sizes, system
interconnects, and integrated memory
controllers.
Scalable shared memory of Intel® QuickPath
technology features memory distributed to each
processor with integrated memory controllers
Figure 2. Intel® Hyper‐Threading Technology
www.advantech.com/NCG
3. Xx
and high‐speed point‐to‐point interconnects to unleash the performance of future versions of
next‐generation Intel® multi‐core processors.
An Integrated Memory Controller supports high‐speed, three‐channel DDR3 memory.
By eliminating the front‐side bus (FSB) found in previous chips and by taking advantage of DDR3
memory the new on‐die memory controller enables higher capacity (to 96GB), higher
performance, and lower power than DDR2‐based systems. The new architecture significantly
reduces memory latency and memory bandwidth is boosted to up to 25.6 GB/s per CPU.
Multi‐level shared cache improves performance and efficiency by reducing latency to frequently
used data.
The Tylersburg‐EP/Nehalem‐EP Platform comprises the Intel® Xeon® Processor 5500 series processors
along with the Intel® 5520 chipset or 36D I/O hub with 36 PCIe 2.0 lanes. The 5500 series has 4 versions;
the E5504 (dual core) and E5540 (quad core) both in 80W packages, and for power efficiency the L5518
and L5508 rated at 60W and 38W respectively, both with quad cores.
All of the Nehalem EP 5500 series processors have extended life support, however the L5518 and L5508
both have High Tcase for constant performance during temperature excursions making them perfect for
NEBs Level 3 and ATCA certifications. The performance, flexibility, low power and packaging of the 5500
series makes them the perfect choice for demanding Storage, Medical, Security, and Communications
Infrastructure applications.
Multi core designs don’t in themselves enable greater performance unless you can feed and manage
them efficiently and optimize software execution to gain every last ounce of benefit from those extra
cores. The Intel® Xeon® 5500 series processors have a number of innovative enhancements that deliver
on the promise of multi core execution.
Greater Parallelism increases the amount of instructions that can be run “out of order.” This enables
more simultaneous processing and overlap latency. To be able to identify more independent
operations that can be run in parallel, Intel increased the size of the out‐of‐order window and
scheduler, giving them a wider window from which to look for these operations. Intel also increased
the size of the other buffers in the core to ensure they wouldn’t become a limiting factor.
Simultaneous multi‐threading (SMT) for enabling a more energy efficient means of increasing
performance for multi‐threaded workloads. The next generation microarchitecture’s SMT capability
enables running two simultaneous threads per core—an amazing eight simultaneous threads per
quad‐core processor and 16 simultaneous threads for dual‐processor quad‐core designs.
Application Targeted Accelerators extend the capabilities of Intel® architecture by adding
performance‐optimized, low‐latency, lower power fixed‐function accelerators on the processor die
to benefit specific applications. There are seven Application Targeted Accelerators included in the
next generation microarchitecture providing new string and text processing instructions to improve
performance of string and text processing operations. For example, they enable parsing of XML
strings and text at a much higher speed. These Application Targeted Accelerators will be useful for
lexing, tokenizing, regular expression evaluation, virus scanning, and intrusion.
www.advantech.com/NCG
4. Xx
EndtoEnd Hardware Virtualization
Intel has made significant enhancements in the area of next‐generation Intel® Virtualization Technology
by adding new hardware‐assist capabilities across all elements in the processing platform:
• In the Processor:
Improvements to Intel® VT‐x provides hardware‐assisted page‐table management, allowing the
guest OS more direct access to the hardware and reducing compute‐intensive software
translation from the VMM or Virtual Machine Monitor. Intel® VT‐x also includes capabilities for
flexible workload migration and performance optimization.
• In the Chipset:
Intel® VT‐d helps speed data movement and eliminates much of the performance overhead by
giving designated virtual machines their own dedicated I/O devices, thus reducing the overhead
of the VMM in managing I/O traffic.
• And last but not least Virtualization has been carried through to the Network Adapter where:
Intel® VT‐c further enhances I/O solutions by integrating extensive hardware assists into the I/O
devices used to connect servers to the core network, storage infrastructure and other external
devices. By performing routing functions to and from virtual machines in dedicated network
silicon, Intel® VT‐c speeds delivery and reduces the load on the Virtual Machine Monitor (VMM)
and server processors, providing up to two times the throughput of non‐hardware‐assisted
devices.
All of this is great news for networking and telecom systems where packet processing takes place over
multiple network interfaces and can take full advantage of end‐to‐end hardware virtualization to
maximize I/O performance.
Getting the most out of 10 GigE
The design enhancements in Intel’s new 82599 10 Gigabit Ethernet controller play a main role in end‐to‐
end network performance and throughput. Compared to the previous generation Intel® 82598, the
Intel® 82599 is packed with new and improved features contributing to up to 2.5 times improvement in
LAN throughput. Advantech’s latest platforms based on the Intel® Xeon®processor 5500 series provide
the architectural elements, such as higher processing power, new local memory architecture, and faster
PCI Express 2.0 I/O interface bus, to enable increased levels of 10GbE scalability. Corresponding
hardware optimizations in the Intel® 82599 include a new PCI Express 2.0 interface (5 Gbps) to improve
the entire data path as well as intelligent queue support (VMDq) optimized for multi‐core processors.
The controller itself is a 6 watt single‐chip, dual‐port 10GbE implementation in a 25x25 mm package. It
helps to reduce BOM cost and design complexity by integrating serial 10GbE PHYs including SFP+
support. Wide internal data paths eliminate performance bottlenecks by efficiently handling large
address and data words. The controller also includes advanced interrupt‐handling features and uses
efficient ring‐buffer descriptor data structures, with up to 64 packet descriptors. A large on‐chip packet
buffer maintains superior performance.
www.advantech.com/NCG
5. Xx
With industry‐leading power consumption, a small footprint, and integrated PHYs, the controller is
ideally suited for ATCA blades like the Advantech MIC‐5320 and for mezzanine card implementations
used on Advantech’s FWA‐6500.
The Intel® Tylersburg‐EP/Nehalem‐EP Platform enables the next step up for architects, engineers and
product marketers alike. Through the utilization of processors bridging multiple generations,
performance levels and price points, building a compatible and scalable full range solution becomes
even easier. Advantech can supply pre‐packaged compact servers and ATCA blades covering the full
spectrum, leveraging the full advantage of software compatibility and greater scalability.
Advantech products
MIC5320 AdvancedTCA 1GigE / 10GigE CPU Blade with Intel® Xeon® 5500 Series Processor
Advantech’s MIC‐5320 single‐slot AdvancedTCA® processor blade
combines computing performance with I/O flexibility in a power
efficient design. Supporting Intel’s® latest Xeon® 5500 series
processors using the Nehalem architecture and latest DDR3
technology with a 3 channel memory controller integrated into the
CPU, the MIC‐5320 outperforms previous generation dual socket
designs. Even with such an increase in power the MIC‐5320 has
excellent thermal characteristics. With built in Westmere readiness,
there is a smooth upgrade path to future processors that are able to
support more than 4 cores.
Using Intel’s latest GbE and 10GE MAC solutions supporting enhanced
offloading techniques and virtualization features users are able to
deploy the full power of this innovative multi‐core technology. The
MIC‐5320 is a versatile solution, combining the most powerful multi
core technology, low latency/high speed DDR3 memory subsystem
and latest 10GE technology, this AdvancedTCA Blade is well suited for
high speed data plane applications. Supporting up to 48GB of
Figure 3. MIC‐5320 ATCA Blade
memory this blade can run database in memory applications easily ‐
backed up by a 4 channel SAS RAID controller that makes the blade equally suitable for control plane
applications that require disk IO with RAID and failover support.
The mid size AMC bay supports more than just mass storage AMCs. With support for PCIe x4 gen 2 as
well as base fabric channels and telco clocks interfaces, it opens up possibilities for high speed IO
interface integration, telco backhaul interface modules and co‐processing engines. In addition to
utilizing the chipset’s RASUM features, redundant firmware hubs and onboard USB disks, BIOS and
firmware enhancements to support CMOS backup, override and HPM.1 upgradeability make this blade a
true carrier grade solution.
www.advantech.com/NCG
6. Xx
MIC‐5320’s overall design and built in flexibility using FPGA technology and RTM customization
broadens the application fields for this product and reduces time to market. The dual processor version
(MIC‐5322) targets specialized applications where enhanced features are required. Advantech’s world
class customization services are ready to tune the blades to customer specific requirements. The
standard features of the MIC‐5320 include:
One Quad Core Intel® Xeon® Processor 5500 Series (Westmere Ready)
Intel® 5520 / ICH10R server class chipset
6 DDR3 VLP DIMMs up to 48GB with ECC support
Two XAUI ports on Fabric interface (Intel® 82599)
Two 1000Base‐Tx ports on Base interface (Intel® 82576)
Three 1000Base‐BT front panel ports (Intel® 82599)
One mid size AMC slot with SAS/PCIe/RTM/CLK support
Onboard Serial Attached SCSI (SAS) controller with failover support
fully managed, hot swappable RTM
FWA6500 Network Application Platform with Dual Quad Core Xeon®5500 Series
Processors
The FWA‐6500 is an innovative and versatile multi‐gigabit network application platform based on similar
core design blueprints as Advantech’s MIC‐5320 AdvancedTCA (ATCA) Blade. It comes with two quad‐
core Intel® Xeon® Processor 5500’s in an enterprise server form factor aimed at the next wave of high
performance networking applications.
Network connectivity is fast and
flexible with up to 16 front
accessible GE ports supported by
4 quad Ethernet modules based
on the Intel® 82576 or with
multiple dual 10GE modules
based on the brand new Intel®
82599 dual 10 GbE controller.
The modules plug in to a 32‐way
PCI Express mid‐plane providing
the high speed interconnects to
the IO controller hub which makes
the system fly. Both RJ45 or SFP
Figure 4. FWA‐6500
based modules can be supported
and can be mixed and matched as required. Two further PCIe x4 slots are available internally for
standard add‐in cards for offload purposes or NPU co‐processing.
www.advantech.com/NCG
7. Xx
The combination of the latest Intel® processors, chipset and Ethernet controllers in one platform
providing acceleration and off‐
loading features gives customers the
combination of performance and IO
scalability they have been waiting
for. The system incorporates up to
96GB of DDR3 memory making it
perfect for extreme speed,
persistent, in‐memory database
applications where performance and
reliability are key. To increase the
FWA‐6500’s flexibility and fault
tolerance level, two swappable SATA
hard drives and a 2U redundant
power supply are integrated into the
base model. An internal remote
management board provides Serial
Figure 5. FWA‐6500 Core Architecture with GigE & 10 GigE modules
and USB connectivity as well as two
LAN ports for remote & out of band
management. IPMI 2.0 is supported along with remote monitoring and power cycling and the ability to
access the console via the LAN (SoL). The system has been designed with NEBS in mind to address more
stringent telecom environments extending beyond the enterprise and data center.
Software Choice
To get the most out of a processing system based on the Tylersburg‐EP/Nehalem‐EP Platform, software
must play its part and be capable of making the most of these multi core engines. Advantech has a
proven track record of building strategic partner relationships with the existing core software providers
as well as the new emerging players to ensure their customer can obtain the best combination of
hardware and software. As with the existing range of flexible computing and communications
application platforms and high performance blades Advantech will make available a variety of industry
leading operating system choices and application enabling software for server and carrier grade
applications.
Summary
With the introduction of the Advantech MIC‐5320 & FWA‐6500 both supporting the Intel® Nehalem
Microarchitecture and Xeon®5500 series processors one can gain access to the latest technology in
multiple form factors – the best of all worlds. Advantech, as a seasoned provider of computing and
communications solutions, is the ideal partner to help create high performance, optimized and cost
efficient platforms for a broad range of today’s applications.
www.advantech.com/NCG
8. Xx
Advantech provides mission critical hardware to the leading telecom and networking equipment
manufacturers. Our products are embedded in OEM equipment that the world’s communications
infrastructure depends upon. With an extensive deployed base, Advantech designs both standard and
customized products for AdvancedTCA, AdvancedMC, MicroTCA and INCA. We team up locally with
customers to evaluate project requirements, share design knowledge and develop optimized solutions
together—all backed up with global deployment expertise with logistics and integration centers on all
major continents.
For more information on these products and Advantech’s range of communications and computing
solutions visit http://www.advantech.com/NCG
www.advantech.com/NCG