Market Mantras

After last night’s star-studded Academy Awards, the time had to come for the nominated to be modest and say that being nominated is an honor in itself, which it is, but let’s face it, winning is better. Like many of last night’s stars, we sat in the audience at the Network Computing awards dinner trying to be cool, but we wanted to win, and we are glad to say that Emulex’s OneConnect Universal Converged Network Adapter (UCNA) was named the Network Computing “Product of the Year” for 2010. We first would like to thank the people who voted for the OneConnect product. We really think this shows how the message of network convergence has reached and resonated with the IT world. As we have stated many times, the road to network convergence will be a long one with many stages of deployment over the next three to five years. However, the industry, vendors, OEMs and entire ecosystems have to line up behind a technology before it can reach a point of deployment and trust.

This past week, Emulex also announced new design wins with HP (click here) and IBM (click here). These design wins are at the core of the strategic movement by OEMs to 10Gb Ethernet (10GbE) for network convergence. IBM’s Virtual Fabric, HP FlexFabric and Cisco’s UCS are driving next-generation computing fabrics and are all based on 10GbE for IP, iSCSI and Fibre Channel over Ethernet. This award is another proof point that Emulex is leading the convergence market and that the OEMs are moving the world toward 10GbE-based network convergence.

Emulex recently held its sales kick-off and asked Steve Duplessie to regale us with his thoughts about Emulex, the market, IT spending, the cloud and numerous other topics. In his own inimitable way, he shared his rants and raves. Here are some of my favorites. Or at least the ones I can publish on a PG-13-rated blog.

Who wants more than ONE? Only when forced do people chose multiple means of performing the same basic function. Convergence makes sense and it will happen because paying for more than one network if you don’t have to does not make sense.
Read the rest of this entry »

Today, we had the Wikibon FCoE Fact vs. Fiction Peer Insight call with a great panel of industry bloggers and the Wikibon team, including Dave Vellante, Dennis Martin, Stuart Miniman, David Graham and Nigel Poulton, as well as a virtual cast of thousands. The core question of the day was: how do we separate the hype and reality of Fibre Channel over Ethernet (FCoE) for the data center today? In listening to the panel and the guest on the call, I came away with three key lessons:

1. FCoE is coming to market and we better get ready. Yes, Fibre Channel will be strong for a number of years. 2010 is the year to do some pilots, and the Wikibon S-curve says 2011 and 2012 will be the point of obvious transition due to cost savings.
2. Vendors have to do a better job defining the value proposition. As a vendor, we need to tone down some of the hype and provide a clear value proposition in numbers and dollars. We are going to work with Wikibon on a calculator and planning tool to make the value proposition clear.
3. We need to provide a better understanding of how companies can get rid of stuff. Saving cash is not enough: make it simple by getting rid of extra gear. We need to do a better job of explaining what can go and how to transition with minimal fuss and headaches. Consolidation needs to come to networks, just like servers and storage.

I enjoyed the call, and I look forward to seeing how the market matures over the next six months. One thing is certain: we (vendors) don’t know as much as we think. The job of bringing FCoE to market is just beginning. Click here to listen to this discussion through the Wikibon audio archives.

With all of the focus on the host side with the announcement of general availability for our OneConnect Universal Converged Network Adapters (UCNAs) and the industry’s first 16Gb Fibre Channel Host Bus Adapters (HBAs) design, we can’t forget about the other side of the networking equation: the target side. After all, without RAID, tape, virtual tape library (VTL), deduplication and hundreds of other target devices, we wouldn’t have a complete solution. Over the past month, Emulex had announced some key solutions for the target side that include our new TargetConnect product and Quad-Port 8Gb Fibre Channel HBA.

Emulex TargetConnect

The Emulex TargetConnect™ Software Developer Kit (SDK) provides the flexibility to quickly develop and deploy target-side solutions based on Emulex Fibre Channel and Fibre Channel over Ethernet (FCoE) CNA target or initiator mode drivers. The TargetConnect SDK makes it fast and simple to support network and storage appliances, test and diagnostics applications, VTL, RAID, deduplication and more.
Read the rest of this entry »

This week, both Dell‘Oro Group and IT Brand Pulse released their first reports on network convergence market share for Fibre Channel over Ethernet (FCoE) and Emulex is the leader of the market for the first three quarters of this new market triathlon. Yes, triathlon, not marathon, because you can’t just be good at one event here to win. You must be able to swim, bike and run, or, in the IT world, you have be a leader in IP, iSCSI and FCoE to win this race. Most of you who have met me know that I am hardly a tri-athlete, but that does not mean that Emulex does not compete just as hard as these athletes in our own three-event race in the IT world.

Emulex Is at the Top of the Leader Board

According to the numbers, Emulex leads in both revenue, 60.1%, and ports, 70%, over QLogic, Intel and Brocade in this early phase of the market. As we move into the second phase of the market, we know that, OneConnect, our tri-athlete Universal Converged Network Adapter (UCNA), will outdistance the single-event specialists in IP (Intel/Broadcom) or FCoE (QLogic/Brocade). The reason is simple: OneConnect provides full three-protocol hardware offload for IP, iSCSI and FCoE on a single platform; it provides a pay-as-you-go feature to lower 10GbE deployment costs; and it provides these differentiated features universally for servers with Local Area Network on Motherboard (LOM), mezzanine and adapter form factors. If you look at how IBM has implemented their virtual fabric solutions on Emulex OneConnect, they have built a solution that lowers costs and improves performance and flexibility for IT managers. This is the first of many innovations based on our industry-leading technology and implementations.
Read the rest of this entry »

Steve Duplessie, in his own unique way, rightfully said we (Emulex) were tickled with the announcements this week from Cisco, VMware, EMC and HP around how convergence and 10Gb Ethernet (10GbE) will be used to connect the next generation of clouds and the data center on the whole (See: http://www.thebiggertruth.com/2009/11/vce-between-the-lines/).

Obviously, it is great to hear major players in the industry talk about the technology and solutions you sell in a positive light, and even better when those solutions are a core foundation of their key initiatives.

To me, the most significant thing about Steve’s blog is his comment, “The world of IT works like this,” if the big guys announce stuff, the rest of the IT world generally follows. However, these announcements are usually the culmination of years of standards work, many generations of development and many startups plowing the new ground to get major players ready to take a stand publicly. This next step into the virtual data center and clouds services is very interesting because it starts with some big fish talking about it first. Yes, they are using lots of building blocks from their construction kits, but it is a new architecture, even if they use many of the same materials.

So, yes, are we tickled! If you look at VCE, HP FlexFabric and IBM Virtual Fabric Solution, they are virtualized, state-less architectures and use networking to reach storage, the cloud and everything else. The cloud is all about delivering the right class of service to every IT need, and the dynamic provisioning tool inherent in 10Gb Enhanced Ethernet-based network convergence is a perfect complement to this solution. Why? It leverages the four core values of the network convergence model:

Enhanced Ethernet-based networking convergence provides support for the core networking, clustering and storage protocols required to deploy the cloud and lower networking costs.

• 10GbE Virtualized Network Interface Cards (NICs) and Universal Converged Network Adapters (UCNAs) – Deploying networking on virtual machines (VMs) requires the same type of hardware redundancy and network segregation as traditional servers, and the ability to create virtual NICs, iSCSI and Fibre Channel over Ethernet (FCoE) adapters gives IT managers the flexibility required to deploy separate and redundant connections for Local Area Network (LAN) management, IP networking and storage (NAS, iSCSI and FCoE).

• Pay-As-You-Go I/O – One of the key value propositions of the cloud is the ability to pay-as-you-go for resources, and Emulex’s UCNAs provide the same business model for networking to these new architectures.

• Unified Management – Consolidation and dynamic control via unified management of resources for each management domain is provided for applications, LAN and storage with a single security model.

So yes, we are tickled to see how network convergence is being used in the market by our OEM and ecosystem partners. We also think IT managers will see a path forward that protects their current investments.

Always the technology fashionistas, QLogic has added some “bling” to its new 8Gb/s quad-port Fibre Channel Host Bus Adapter (HBA) (QLE2564) – two shiny new heat sinks. “Heat sinks?” we asked. You can only imagine our surprise! For the past 18 months, QLogic has been touting its “Cool HBA Technology.” The theory of this highly touted feature was that QLogic 8Gb/s Fibre Channel HBAs did not need a heat sink for reliable operation, but thermal images, such as those below, show that they really do:

What is most surprising about this sudden shift is that it took QLogic so long to catch up with the design best practices which Emulex and the rest of the industry have been leveraging since the beginning.

The fact remains that heat sinks are vital to achieving enterprise class reliability and are one of the best ways to cool devices in today’s high-density computing environments. As you can see in the images below, it has taken QLogic over 18 months to truly achieve its “Cool HBA Technology,” and even then, only after following Emulex’s lead.

So why is cooler better? It’s simple. Cooler operating adapters mean greater reliability – of significant importance to data center administrators, as properly cooled components have a longer operating life, better system reliability and greater Storage Area Network (SAN) availability, each critical considerations in blade server environments. The cooler operating Emulex 8Gb/s Fibre Channel I/O controller translates into 166% greater reliability when compared to QLogic’s QLE2562 HBA.

The real question QLogic customers and partners have to be asking now is “what about all those QLogic 8Gb/s cards in the field which have no heat sink?”

To learn more, check out the following:

Emulex Labs: Sometimes, Being Hot Isn’t So Cool: http://www.emulex.com/emulexlabs/?p=14

The Effects of Heat on Electronic Circuits and Devices: http://www.youtube.com/EmulexVideo#play/uploads/17/AQOaNX9C6JE

With all the talk about network convergence this week, you might be thinking that we have forgotten about Fibre Channel. No way! This month, the Fibre Channel Industry Association (FCIA) ratified the 16Gb Fibre Channel standard, and this week, we announced the industry’s first 16Gb design win with our long-time partner, IBM p Series.

Fibre Channel Market Muscle

According to the Dell Oro Group, Fibre Channel will be a $675 million market by 2013, so it is not going anywhere. Why? Installed base and investment protection is a key factor, but it goes beyond that. Fibre Channel is one of the proven cornerstones of the data center, and IT managers will not just abandon a proven solution. Fibre Channel will be a dominant storage interconnect as a standalone transport or encapsulated in FCoE for at least another decade.

The Powers of 2

1, 2, 4, 8 and now 16Gb. Fibre Channel continues to move forward, and yes, 32Gb is on the FCIA roadmap as well. Each generation of Fibre Channel has followed a power of 2 and shows no sign of stopping. Here are some details from the press release: ”The Fibre Channel Industry Association (FCIA) ANSI INCITS T11 committee completed the technical work on the FC-PI-5 for 16Gb/s Fibre Channel (“16GFC”) and voted in early October to send the document out for letter ballot. This milestone marks the technical stability and completeness necessary for vendors to commit to silicon their upcoming designs based upon the FC-PI-5 standard…” This is exactly what IBM and Emulex have done with our announcement this week.

As you would expect, 16Gb Fibre Channel users will experience twice the bandwidth of 8Gb Fibre Channel, and like previous generations of Fibre Channel, 16Gb Fibre Channel will auto-negotiate backward compatibly to 8Gb Fibre Channel and 4Gb Fibre Channel. 16Gb Fibre Channel provides a natural migration path from 8/4Gb Fibre Channel and ensures the end-user full confidence that 8Gb Fibre Channel purchases made today are preserved investments for tomorrow.

16Gb Fibre Channel will provide more virtual ports to improved efficiencies with high-density multi-core CPUs, improve single-root hypervisors support, improve IOPS for solid state disk drives and make way for third-generation PCI Express (PCIe) bus architectures and shared I/O deployments. Additional information on the standardization efforts of 16Gb Fibre Channel is available at the ANSI T11 Web site: www.t11.org.

How Do We Reconcile Support for Fibre Channel and Fibre Channel over Ethernet via Network Convergence?

Fibre Channel and Fibre Channel over Ethernet (FCoE) are complementary, not competitive. Fibre Channel is a core part of network convergence. By supporting the latest innovations in Fibre Channel, we are enabling network convergence and advancing tools that will help us build better, more flexible and more powerful FCoE implementations for 10GbE and 40GbE in the future.

Yes, Emulex is a strong advocate of network convergence. We see network convergence and FCoE as a market expansion opportunity, not a market replacement. We started this blog by saying Fibre Channel will be a big market for many years to come, and we plan to keep growing in our core Fibre Channel markets.

This week, we announced General Availability of the OneConnect Universal Converged Network Adapter (UCNA) and OneCommand Manager, our network convergence management framework.

Today’s blog is not about the product, but rather the launch process, contrasted to the real season, when games count. I know we are supposed to be a global company, and I should be probably be using analogies to what the rest of the world considers to be football (soccer) versus NFL football, but it is NFL football season here in the U.S., so please bear with me.

Training Camp

Product launches are much like getting ready for the NFL season. You hope that the players have worked out during the off-season to be ready for training camp. If they are not, then you put them on a rigorous get-in-shape regimen and drive them hard to get ready for the season. You look at the moves other teams (competitors) have made, new strategies they are likely to implement and how they will defend against you. When they attack your company, your products, your history, your future, it is all fair game during the season. Just ask the Cowboys’ Tony Romo– I bet he wishes he never dated Jessica Simpson or brought her to a game!

The Preseason with Analysts and Press

The preseason is the right time to wring out the game plan and see if your story, products and messages resonate or get ridiculed. Sometimes this is a humbling process, and often it is a fun one, but it is always educational. You learn quickly what puts people to sleep, what they find intriguing and what you wish you never said. This is a vital part of the launch process, because it really helps separate the signal from the noise.

The Launch Event

The stands are filled and the crowd is roaring. The ball is about to be kicked off and you are on stage, or more likely, webcast, these days. The team is tense, tired and excited, all at the same time. The months of nights, weekends and long hours are about to be put to use. For the OneConnect and OneCommand launch, we kicked off in a big way with three global events on three continents in 24 hours, starting in Frankfurt at SNW, in New York at our investor relations analyst day and in Beijing at our channel partner conference.

Trash Talk: Time to Put Up

Trash talk is essential and integral to competition in every game, and there has been plenty of that leading up to this launch. We all jockey for position and leadership in the market, but that time has passed and it is time to stop talking and start playing.

Trash Talk #1: Emulex is late to market.

Game Day Reality: We are right where we want to be, on target for the major 10GbE server transition.

Trash Talk #2: The product does not exist.

Game Day Reality: You can buy it now in the channels and from IBM. Did they think this was really a competitive strategy, since we qualify with the same OEMs? Our competitors knew this was wrong.

Trash Talk #3: They don’t own the chip.

Game Day Reality: The world’s biggest OEMs have qualified our OneConnect product for 10Gb network interface card (NIC), iSCSI and Fibre Channel over Ethernet (FCoE) uses (see the IBM Web site). The only people that seem to have a problem with our product are our competitors. Who do you want to listen to? Besides you really didn’t expect them to say nice things about us, right?

For me, the reality check is to look at the “quality of the trash talking.” If the only things our competitors can talk trash about have nothing to do with the product features, capabilities or functions, they can trash talk all they want. To me, this means that they don’t have competitive advantages where it counts—in their products or business model. Maybe their products are trash, and all they have is talk?

It’s the Final Score that Counts

We have said this many times: the network convergence market is not going to happen overnight. This year has been the year of OEM qualifications, and 2010 will be the year of network convergence program pilots and budgeting, and the market transition will kick off with production deployments as we move into 2011. The pre-season is over and the real game begins now. We won’t know the final score for a while. It is time to start working for the market share playoffs. In 2006, when asked if his Pittsburgh Steelers could win the Super Bowl, coach Bill Cohwer replied, “I like our chances.” That is how I feel about the network convergence market. We are prepared, we have had a great pre-season, we have the right product, at the right time, with the right partners and the best business model. So yeah, I like our chances.

We recently announced that we had published a new set of industry certifications with Ixia, the Finisar of the IP world for all of you storage folks. The test results are called the “Black Book.” IBM has the “Red Book,” and I am sure that there are other color-coded books out there. However, this is significant because it is the first non-OEM derived third-party testing and validation of our new 10GbE OneConnect Universal Converged Network Adapters (UCNAs). It is also the first time a test equipment manufacturer from the Ethernet world has publicly tested and supported a new 10GbE converged network adapter.

Share and Share Alike

The joint venture also was Ixia’s first venture into the storage world, and we were able to help them develop a testing suite that not only validated our 10GbE protocol capabilities, but also extended their product to help test IOPS and bandwidth in new ways required by today’s storage administrators and vendors. In turn, they helped us see the market and networking through the eyes of an IP administrator. Just like in kindergarten, learning to share was the first lesson in building a better solution for our respective markets.

What Is in the Ixia Black Book?

A primer on the technology and testing of CNAs, the Black Book begins with an introduction to CNA technology, the developments that led to the Fibre Channel over Ethernet (FCoE) protocol and data center network convergence. It includes detailed test plans using Ixia’s leading benchmark tool, IxChariot™, which guides the reader through key tests required to thoroughly measure the performance of a CNA. Step-by-step instructions include measurements for Transmission Control Protocol (TCP) throughput, User Diagram Protocol (UDP) throughput, latency and storage input/output (I/O) performance in virtualized and non-virtualized environments. (Download the Black Book from Ixia’s Web site)

The book outlines the basics of network convergence, what makes Enhanced Ethernet special and how it can be used to support IP, FCoE, RDMA, NAS and iSCSI on a single network. Next, it reviews IxChariot, the IP industry’s leading benchmark tool for measuring network interface card (NIC) performance since 10Mbps Ethernet adapters were first introduced on PC platforms. We have learned through this partnership how to use the advanced scripting language to test every kind of endpoint in your network:

• Basic Throughput – Using the simplest methodology for measuring TCP throughput on networks and devices to establish a TCP connection from Endpoint 1 (E1) to Endpoint 2 (E2). Then test performance using the system default of 8KB to 32KB per block.

• High-performance Throughput – the high-performance throughput script may be used to enable Winsock’s overlapped I/O mechanism to increase the efficiency of network transactions by allowing an application to queue multiple requests. The high-performance throughput script has a default file size of 10MB and a default send buffer size of 64KB. This script is ideal for testing TCP throughput on 100Mbps, 802.11n and 1Gbps networks.

• Ultra-high Performance Throughput – The ultra-high performance throughput script is designed for testing in the world of 10Gbps Ethernet and beyond. This script uses overlapped/asynchronous I/O as in the high-performance throughput script, but increases the default settings for socket buffers, file sizes and send buffer sizes in order to maximize the benefit of TCP offloading technologies.

• UDP Throughput – The UDP throughput script was added in IxChariot 7.0 in order to provide an optimized method for measuring peak UDP throughput on the network. This script will stream UDP datagrams from E1 to E2 as quickly as the sender can put them on the network. There are a few key items that should be noted for UDP throughput testing.

• Latency – The IxChariot response time script implements a classic ping-pong measurement of machine-to-machine roundtrip time. E1 sends 100 bytes of data in a single small packet to E2, which immediately responds with another 100-byte packet back to E1. By measuring the time it takes to send a few thousand ping-pong transactions, the average roundtrip latency between two endpoints can be estimated.

• Application Performance – IxChariot includes more than 100 scripts based on modern Internet and enterprise network applications. These scripts can be used to measure the typical performance of these applications over a network between any E1 and E2. For example, customers who are building application servers will use application scripts matching their expected profile to understand how a specific server and network adapter combination will perform on their network. Database applications are a prime example in this category, since they are so heavily transactional in nature; a minor difference in overall system latency can translate into large amounts of idle time observed by end-users.

• VoIP and Video – Finally, IxChariot has unique simulation capabilities for Voice over Internet Protocol (VoIP) and video traffic. IxChariot was the first tool on the market to implement accurate measurements of call quality based on the ITU G.107 E-Model specification. To measure the impact of network quality on call scores, IxChariot uses its UDP/RTP streaming capabilities. It sends simulated media frames in order to measure the jitter, latency and loss, which play a significant role in users’ experience.

The tests have a dedicated section on testing done on virtual machine (VM) performance in server virtualization environments. Needless to say, the IxChariot testing covers the gambit of performance scenarios and provides a solid basis to evaluate the OneConnect UCNA and how it will help your 10GbE network performance today and in preparing for network convergence in the future.

Picking the Right NIC

In my last blog, we discussed the importance of picking the right NIC to enable network convergence. The Ixia Black Book testing results are a good indication that the Emulex OneConnect UCNA is the right NIC to start your network convergence transition.