Click graph to expand: See where Broadcom ranks with other chipmakers.

The third time is definitely the charm for Broadcom’s Ethernet Controller group.

We’ve again been lauded by IT Brand Pulse, an independent market research firm that conducts surveys to determine end-user perceptions of the top industry vendors. They look at overall market leadership, performance, innovation, reliability, service/support and price.

In the IT Brand Pulse 2013 Brand Leadership Survey, Broadcom not only won the Ethernet Local Area Network-on-Motherboard (LOM) Market Leadership award for the third straight year, we also won the Performance, Innovation and Price leadership categories.

The fact that this survey is completed by the actual users of the technology products, rather than a hand-picked board of vendors or academics, is a really big deal for the hardworking engineers at Broadcom. There are myriad industry awards, but this is the only independent survey that provides a comprehensive view of customer perceptions.

We’re excited to have been selected as the brand leader for our third consecutive year, particularly by the information technologists who depend on our products to run their businesses each and every day.

I’m very proud of our dedicated employees for their commitment and hard work, but also our customers for their continued support and recognition of our award-winning NetXtreme network technologies for the data center.

Related:

• Broadcom Wins Market Leadership Award for Second Year in a Row!
• Going NetXtreme: Dell and HP tap Broadcom for enhanced server technology

Click to expand. Source: Microsoft Corp.

Unfortunately, I cannot take credit for the phrase, “the whole is greater than the sum of the parts.”  This is generally attributed to Aristotle – admittedly, a tough act to follow – but I will borrow from it as a way of highlighting my thoughts in this blog post.

With the highly anticipated release of Microsoft Windows Server 2012, the industry is anxious to see what new features will have the greatest impact on users. While this is important, I would suggest that it is more important to look at the overall impact the combined technological advancements will have on the industry.

Read Broadcom’s press release here.

Earlier this year, we saw several important industry announcements from major server vendors promoting Xeon E5 “Romley” – based platforms. As part of these announcements, Broadcom released new 1GbE and 10GbE NetXtreme Converged Network Adapters supporting L2 networking and iSCSI/FCoE storage offload protocols. This was critical because it is important to maintain “balanced” performance in a system. There is limited benefit in improving the processing capability of the server if the overall performance is bottle-necked by older I/O technology.  By increasing both the processing power and the server I/O, the overall performance of the server is improved.

We are now facing another transition in the industry where three major components of the server are being advanced: The combination of Microsoft’s powerful new Windows Server 2012 operating system, the new Romley-based servers from tier-1 vendors such as Dell and HP, and Broadcom’s new NetXtreme CNA adapters. These components are simultaneously advancing the: operating system, CPU and overall server architecture, and Network and Storage I/O.

While the individual technology improvements are impressive, it’s the combined advanced features and tight integration that makes the “value greater than the sum of the parts.” The work behind the announcements this week began years ago and would never have been possible without the collaboration between these industry leaders.

And let’s not forget the larger benefit of the key components that have been designed to work together.  At the risk of getting too philosophical, I will stop here. I am sure Aristotle would be proud …

To learn more about the combination of Microsoft’s Windows Server 2012 operating system, Romley-based servers and Broadcom’s NetXtreme adapters, download the Enterprise Strategy Group’s white paper: “Broadcom’s Integrated Solutions for Cloud Scale Computing.”

Related:

• Why Being Fast, Fat and Flat Can Be a Good Thing!
• The Road Less Traveled: Server Virtualization and the Data Center
• VMworld Preview: How the Cloud is Reshaping Data Centers
• Going NetXtreme: Dell and HP tap Broadcom for enhanced server technology

Server architecture has come a long way since the days when each server was assigned to a specific application or task. Because many tasks don’t play well with others, each task required its own dedicated machine. It was a very basic, albeit not too efficient, approach to architect a data center network.

With today’s mega data centers and cloud-based services on the rise, computer networks are becoming larger and more complex, rendering the dedicated servers of yesteryear obsolete.  As the sheer number of servers continues to increase, one can imagine the amount of physical space required by a data center that’s not only overcrowded with racks of servers but also consuming massive amounts of power and generating heat.

To move past traditional data center architecture, networking professionals often face a labor-intensive and time-consuming process of reconfiguring workloads. Server virtualization attempts to address both of these issues in one fell swoop while creating more flexibility and greater efficiencies.  Utilizing specially designed software, an administrator can convert one physical server into multiple virtual machines. Each virtual server acts like a unique physical device that is capable of running its own operating system (OS). Virtualization presents many advantages, enabling users to consolidate computing hardware resources and allowing them to run multiple virtual machines concurrently on consolidated hardware.

Now, IT and data center administrators are leveraging encapsulation and “tunneling” strategies to address the networking problems created by complex virtual environments and the difficulties created by extending network segments long-distance between data centers. In essence, tunneling fully abstracts the physical network, extending the VLAN construct to offer multi-data center network scalability.

Broadcom’s NetXtreme II™ has perfected the art of tunneling. Like most network protocol processing, the bulk of the intelligence is done in the NIC in order to ensure low CPU utilization and maximum performance. By eliminating islands and taking advantage of NIC partitioning, Broadcom’s 57810 allows servers to support Network Virtualization Offload (NVO) services with the various Layer 2 over Layer 3  encapsulation formats that not only reduce system power and cut cooling costs but also minimizes overall system complexity. Most importantly, NVO creates better performance with a lower total cost of ownership — the poster child for fast, fat and flat networks in saving time and money.

To see Broadcom’s 10G tunnel offload technology for software-defined network environments, stop by our booth at Interop or visit our newsroom here.

Full Coverage: Broadcom at Interop 2012

• Broadcom at Interop: Power Consumption Technology Plays Important Role
• Broadcom at Interop: Energy Efficient Ethernet is Good for the Planet
• Technology Moving at the Speed of Life: Broadcom Enables Massive Network Scalability
• Enterprise 2.0: Broadcom puts Network Managers in the Fast Lane
• Broadcom at Interop: Next-Generation Data Centers Shift into High Gear
• Broadcom at Interop: Unprecedented Innovation for Next Gen Data Center, Green IT and Enterprise 2.0
• Gallery: The sights of Broadcom at Interop
• 5GWiFi, NetXtreme Ethernet win top awards at Interop

There’s a lot discussion these days among network IT professionals about making the leap from 1 gigabit Ethernet (GbE) to 10GbE. In fact, I recently participated in a Wikibon Peer Incite discussion about this very topic. In a conversation titled “The Rise of 10Gb Ethernet and the Impact of Intel’s Xeon E5 Family of Processors,” we not only talked about the speed advantages of 10GbE but also some of the solutions (such as HP’s Flex-LOM architecture) that make it easier to determine when the time is right to upgrade to faster speeds.

Graphic: Click for detailed view

We also discussed how 10GbE upgrades are having a direct impact on the way data centers are being constructed. The traditional data center has three tiers: presentation, application, and database. This architecture leaves server resources in islands that are optimized for “North-South” networking traffic only. This means that each tier talks to its adjacent tier, but not to servers in the same tier (commonly referred to as “East-West traffic”).

Reconfiguring workloads and adapting to changes in this kind of data center can be very labor-intensive and time consuming. It can also mean having less flexibility and incurring more operating expenses when it’s time to make changes.

To address these challenges, we talked about how the three-tier data center is giving way to the construction of virtual data centers – a structure in which all of the servers are fully connected via 10GbE in a “flat” network (one with fewer tiers). This not only simplifies network construction but allows any set of servers to be configured in any “logical” tier, as the need arises.

Ultimately, this eliminates compute islands, increases flexibility and reduces operating expenses.  This also underscores one of the biggest lessons learned from utility/public cloud computing architectural practices, that fast, fat, and flat networks save time and money. Broadcom is an innovator in this area and has an extensive portfolio of Ethernet solutions and other devices that enable fast, fat, and flat networks.

Click to learn more about Broadcom’s NetXtreme Ethernet Solutions.

Related News: Broadcom Launches Portfolio of Best-in-Class 1G and 10G Ethernet Solutions for HP ProLiant Gen8 Servers

It’s been a busy week at Broadcom and the server industry at large with Broadcom making major announcements around advanced Ethernet technology solutions for both Dell’s and HP’s new line of servers.

The first announcement was a new portfolio of best-in-class 1GbE and 10GbE Ethernet adapters for Dell’s 12th-Generation PowerEdge servers. These adapters feature Broadcom’s latest NetXtreme controllers, as well as the highest performance networking silicon, and allows Dell to boost performance and further improve efficiency and scalability of enterprise servers. (See video below.)

The second announcement was to introduce a new portfolio of adapters and LOM (LAN on Motherboard) for HP’s new ProLiant Gen8 servers. Broadcom’s five new adapters for the HP line are available in a variety of dual-port and quad-port network interface cards (NICs), as well as FlexNet cards and mezzanine cards. They are designed for a wide range of network configurations and provide full compatibility with HP’s new servers.

Broadcom’s new technology enables IT managers to maximize performance, efficiency and scalability in industry news servers from Dell and HP. The performance delivered is with industry-leading speed and full line-rate throughput across all ports.

Our new 10GbE Ethernet adapters deliver up to 37 percent faster throughput than the nearest competitor. Consider the results of these third-party benchmark tests conducted by Demartek that compared Broadcom’s adapter to the equivalent adapter offered by Emulex.

In addition, these adapters provide full HBA-level offload support of iSCSI and FCoE storage protocols that drastically reduce CPU and memory bandwidth utilization while delivering up to 1.5 Million iSCSI I/Os per second (IOPS): 120 percent higher than the nearest competitor!

The adapters also offer advanced functionality, like switch-independent NIC partitioning (for I/O virtualization) and IEEE 1588 time synchronization. They also implement Energy Efficient Ethernet (EEE) technology that enables ports to use up to 42 percent less power while lowering  IT operating costs.

As I mentioned in the beginning of this post, this has been a busy week of announcements and we are happy to unveil our new LOM and Adapter OEM products to the industry.

In my last blog post, I talked about the upcoming launch of Romley/Sandy Bridge servers and how network partitioning is an easy and cost effective way to help maximize the value of those new solutions. This week I’d like to expand further on the topic of maximizing value and ROI for new servers while sharing some key insights that I gathered during last week’s Ethernet Technology Summit in San Jose.

For those of you who aren’t familiar with this event, the Ethernet Technology Summit is a three-day conference focused on the latest advancements in Ethernet technology (especially the migration from 1GbE to 10GbE, the emergence of 40/100 GbE, convergence, cloud computing, and virtualization). The conference features a wide range of keynote addresses, tutorials and panel discussions from experts who are heavily involved with product development, creating and implementing industry specifications, and of course,  key decision makers who are helping to shape the future of the industry.

Even though the terms “Romley” and “Sandy Bridge” weren’t mentioned heavily in the sessions I attended, it was still very clear that there are several underlying trends and dynamics in the market that are placing a spotlight on this next-generation of server technology. In particular, many questions were being raised about how future solutions will enable managers in the data center and enterprise networks to dramatically increase bandwidth and speed on their servers while simultaneously saving time, money and power consumption. There also was a lot of attention on storage (FCoE/iSCSI), security as well as the migration of 1GbE to 10GbE and how higher bandwidth platforms are on the verge of massive expansion over the next few years.  I noticed a lot of attention being placed on how the transition to new devices and higher levels of bandwidth present many new challenges and opportunities for the industry.

While I attended the conference, I had the pleasure of delivering a keynote address about “Ethernet-based Data Centers in the Cloud Computing Era.” In this session, I discussed the multiple pressures re-shaping today’s data center, including:
• Server and Input/Output (I/O) Virtualization
• Virtual Networks
• Storage and Networking Convergence
• Efficiency (deployment, re-configuration and power)

The new data center is being dramatically re-shaped by these trends and I’m happy to say that Ethernet is the common thread that unites these technologies. Whether we’re talking about the convergence of networking and storage protocols or I/O virtualization inside the server, Ethernet is at the center of these new deployment models. The efficiency and rapid deployment of resources in the new data center is largely due to higher speed networks (faster, fatter and flatter) all the way from their core to the edges inside the servers (advanced Ethernet controllers).

In general, I thought this year’s conference delivered valuable perspectives from some very interesting and influential professionals across the industry.  But again, an underlying theme that was prevalent in many of these discussions revolved around new technologies and how those advancements will be critical in overcoming challenges that so many network and storage administrators face in the not-so-distant future (efficiency in purchasing and running their data centers: Capital Expenditures and Operating Expenditures). That being said, future generation servers are now at our doorstep – primarily Romley/Sandy Bridge technology – and having the ability to maximize the ROI on these systems will certainly be front and center for a very large and growing number of professionals.

For more information about the 2012 Ethernet Technology Summit, visit the Summit’s website. 

In last week’s blog post, I provided Broadcom’s perspective on the soon-to-be-released Romley-based servers and what it means for the industry.

But before I get started, I want to answer a question about “Romley” and how it relates to Intel’s new Sandy Bridge-EP processor (also on the verge of release).  To set the record straight, Romley is actually Intel’s code name for the server platform that employs the Sandy Bridge-EP CPU, so we tend to use them somewhat interchangeably.

In this week’s post, I’d like to focus on another key technology that’s being deployed by network managers and will continue to be instrumental during the Romley cycle. The technology is Network Interface Controller (NIC) partitioning or what is commonly referred to as NPAR.

NPAR is very important because it gives administrators the ability to divide a single “fat” network port into as many as four logical ports and allocate the “fat-pipe” bandwidth into whatever configurations best fit the application.  The end benefit is a better allocation of server resources and better management of those resources, both of which contribute to lowering infrastructure and operating costs.

About a year ago, Broadcom introduced a Network Daughter Card (NDC) that delivers either quad-port 1GbE or dual-port 10GbE.  The  BCM57712-k adapter suddenly gave network administrators the ability to deploy 10GbE on two physical ports or divide the bandwidth into multiple logical ports (up to four per physical port or eight per controller). It’s a very effective solution that tailors network bandwidth to the need at hand.  It also allows for dynamic reconfiguration of that bandwidth. NPAR continues to expand into network systems worldwide.

Some of you may be asking, “but do I also have to buy new and expensive NPAR-enabled switches to use that feature?”

The answer is a resounding “no.” The NPAR feature of Broadcom’s 57712-k is switch-independent, it works with ANY Ethernet switch!  That means you can use Broadcom’s 10GbE adapters along with your existing fabric switch and pass-through modules alike.

Another benefit of NPAR is that it offers a variety of configurable bandwidth controls that allow each logical port to be configured to limit bandwidth to a maximum, guarantee a certain minimum or allow unused bandwidth to be used on a prioritized basis. Bandwidth allocation is very granular, allowing it to be set in 100Mb/s increments.  This level of granularity and flexibility of controls are important because it guarantees bandwidth and quality of service (QoS) for the individual workloads that share a 10GbE port. With NPAR, you can tailor bandwidth based on the workload instead of taking a “one size fits all” approach.

So if you’re upgrading your existing Dell M710HD or M915 blade server, or waiting for the new Romley-based servers to hit the market, be sure to configure it with an NPAR-enabled NDC.

Read more about NIC partitioning (NPAR) and Broadcom’s BCM57712-k Network Daughter Card (NDC).

If you’ve been following recent trends in the server market, you must be as excited as we are here at Broadcom about the upcoming roll out of Intel’s Romley processors, and the subsequent roll out of Romley-based products from server OEMs.   

As the volume of network traffic, networked devices and huge amounts of data continue to ramp at alarming rates, current server and I/O architectures are under the gun to deliver.

With increasing demands for server virtualization, cloud computing, and I/O intensive applications, the need for greater processing capabilities in enterprise networks and data centers is paramount.

New technologies being deployed to meet these needs include 10 Gigabit Ethernet (10GbE) server networking solutions; NIC Partitioning, Data Center Bridging/Lossless Ethernet, Fibre Channel-over Ethernet (FCoE), iSCSI (over Lossless Ethernet), and last but certainly not least, Modular/Flexible LAN-on-Motherboard (LOM)/Daughter Cards.

The trend toward server virtualization has resulted in the deployment of multiple network connections per server. Given the diversity of traffic that a converged 10GbE network can now handle, some 10GbE adapters, including Broadcom’s (shameless plug #1), offer NIC partitioning or separate virtual fabrics that enable a single 10GbE port to be divided into a number of separate ports. Each of these partitions needs multiple queues behind them that allow the controller to steer traffic into separate queues based on Ethernet addresses and other header information (i.e. a hardware-managed 4-Tupple HASH, for those of you who really like details). This is imperative to spread traffic processing across the many processor cores that will soon be available with Romley-based servers. It’s also a unique capability found on Broadcom 10GbE controllers (shameless plug #2).

With the introduction of Romley-based servers over the next few months, we look forward to seeing Broadcom networking technology widely deployed – aligning high-performance Romley processing with high-performance networking and storage.

Want to learn more?

Read this report on Server I/O Considerations, recently published by Crehan Research for Broadcom.

Check out this EE Times story on how Intel’s Romley is Poised to Roil the Data Center.

It’s always nice to be recognized with an award, but it means even more when it’s based on feedback from the folks that actually use our products.

Each year IT Brand Pulse, an independent market research firm, conducts a survey to determine end-user perceptions of the top industry vendors. They look at overall market leadership, performance, innovation, reliability, support and price.

This is incredibly valuable for vendors like us, because there is no other independent survey that provides a comprehensive view of customer perceptions. At the end of the day, there is no one more important than our customers. So, this award is really our “report card” of sorts.

When the results were released this month, we were thrilled to see that Broadcom took the Market Leader award in LAN-on-Motherboard (LOM) for the second straight year, and swept five of the six categories of brand leadership, including market leader, price, performance, reliability and innovation.

Interestingly, in a previous IT Brand Pulse survey, they found that a full two-thirds of server administrators know what brand of LOM device is in their servers because it affects the way their network operates. Customers truly care what technologies are “under-the-hood” running their IT infrastructures.

Needless to say, Broadcom is honored to have been selected as the brand leader, particularly by the IT professionals who rely on our technology every day. This award is a tribute to our continued commitment to providing the best LOM solutions on the market. As we look to the next major data center upgrade with Romley/Sandy Bridge servers, we are excited to be working with our OEM partners to bring the highest performance networking solutions to the market.

Read more about how Broadcom stacked up in a press release announcing the IT Brand Leader Awards (PDF).

So, I would like to thank our employees for their contributions and hard work, but also our customers for their continued support and recognition of our products. Learn more about Broadcom’s Ethernet Controller products.