Intel today released open source code for encapsulating Fiber Channel packets in Ethernet Frames which can be then transported over 10 Gig Ethernet. The software designed for linux is available on Open-FCoE.org. I expect that this will bring in more 10 Gig Ethernet stuff in SANs and data center environments where Fiber Channel has been pretty dominant till now.
Sunday, December 16, 2007
In continuation with the previous post, I will be posting some interesting video clips related to the discussion topics of this forum, to illustrate the growing popularity of this media. Most of these video clips are embedded from sites such as Youtube, Google Video, Revver etc.
Today's clip features a discussion with an NTT official about why we are witnessing adoption of Ethernet PON technologies in Japan while Europe and North America are going with GPON.
Posted by Amitabha at 9:32 PM
Friday, December 14, 2007
Slightly off-topic here... owing to my spare time in the holiday season. I am impressed at how predominant video clips are becoming over the internet... the success of You Tube being just one example. Advertising is increasingly shifting to using video clips over the internet. Look through YouTube and you will find ad clips from well recognized companies including Sony Playstation. As an example, here is Sun Microsystems' video clip advertising its 10 Gig Ethernet adapters... something on topic.
In case you are a video producer or amateur and would like to explore ways of making money over the Internet using your video/ movie clips, here is an excellent article by a well-known independent cinema veteran.
Btw, VMware released their ESX 3.5 with support for Neterion's cards with a claim of delivering 10 Gbps from a virtualized domain with near native OS CPU utilization. Anybody able to verify this?
Posted by Amitabha at 9:59 PM
Friday, November 23, 2007
Just read an article from Nyquist Capital about Google designing its own 10 GigE switches. Its interesting how the authors traced the source of thousand of SFP+ components to Google to determine this. The article discusses how the strategy is very similar to Google designing its own servers and compute farms from basic components as opposed to buying servers from companies like IBM, HP SUN etc.. The article also mentions Google adding 5K+ 10GigE ports a month to manage its 500,000+ compute nodes.
Thats a ton of money being saved, given the cost of 10GigE switches in the market today. Which brings us to the question, what is so proprietary about a switch that vendors like Cisco, Juniper, Woven etc., can sell them at a huge price and a huge margins. Does IOS-X from Cisco have some patented stuff that open source software cannot implement or don't have. I would love to see the above Google technique revolutionize the switch market and motivate some startup to design a switch fabric based on commodity hardware.
Posted by Amitabha at 6:33 AM
Wednesday, October 03, 2007
A bit of shift to the optical technologies in my blog today, with LightReading's Optical Expo going on in Dallas. One interesting thing that strikes me is that today optical stocks are soaring, both Ciena (CIEN) and Infinera (INFN) have > 5% gains with no extraordinary news from either company. May be Optical Expo is delivering a strong message which is going unnoticed. Any coments?
From news reports, 100 Gig technologies are getting quite some attention, with AT&T, Verizon both complaining that the standards may not be ready fast enough. Interesting to note that they have been observing / are anticipating tremendous bandwidth flow in their pipes. There has been much talk about how much bandwidth demand video can hog. Certainly there is extraordinary promise. Netflix is already hosting video and provides download services.
I think more than anything else, it depends on how the access world picks up. Give people optical fiber to their home, give them a 1Gbps bandwidth pipe, and see how much video and bandwidth they will use. The potential is extraordinary.
So my appeal to AT&T, Verizon and the likes. Focus on access. Get all PON technologies rolling fast. Invest your money in digging those trenches to the apartments and homes and you will be extraordinarily rewarded.
By the way, this blog has had double views over the last two days. Is it just me dreaming about the success of Optical Expo?
Posted by Amitabha at 9:48 PM
Wednesday, September 19, 2007
Teknovus, one of the leading companies making EPON chipsets, yesterday announced $28 million funding from a group of equity firms. New investors in Teknovus include Lightspeed Venture Partners and the Galleon Crossover Fund, Vision Venture Capital from Taiwan and Technology Alliance Investment and Itochu Technology Ventures from Japan. Teknovus’ existing investors also participated in this round, including USVP, Partech, Focus, Samsung, and Mitsubishi.
Posted by Amitabha at 7:01 AM
Thursday, September 13, 2007
We turn two years today. For those interested, our first post is here.
It has been a great experience following trends in this very hot and growing field. Just looking back at the progress, Ethernet PON is now widely deployed with 10 Gig standardization coming up, 40/100 Gig standardization is going on, 10 Gig Ethernet adapters are close to entering the volume market, and high speed I/O is now being considered more and more important by chip and server vendors.
Thanks to all readers, all those to supported the blog sending comments and emails. Hope you continue to find this blog useful in the future.
Also we will soon come up with a new layout and a poll. Please help us understand your needs better by participating in the poll. Thanks again.
Posted by Amitabha at 8:00 AM
Tuesday, September 11, 2007
Sipperec , which provides power and telecom services on behalf of 86 municipalities around Paris today announced a fiber-to-the-premises (FTTP) network called Sequantic, which will build on its fiber backbone. This will provide connectivity to buildings targetting about 6,500 enterprises in 147 business areas located in the municipalities. Sipperec will then sell capacity on the network to service providers, so adopting the "open access" model. This is a major boost to this model promoting neutrality to the network provider also adopted by UTOPIA in Utah and the City of Amsterdam.
At UC Davis Networks Research group, we proposed advanced network protocols that may be useful to promote the "open access" model. This protocol aims to guarantee established Service Level Agreements (SLAs) to both users and service providers simultaneously. Our work was published in the IEEE Journal of Selected Areas in Communication, a copy available here.
Posted by Amitabha at 7:58 PM
Monday, September 10, 2007
Matisse Networks which is building Ethernet over Optical Burst Switch technology today announced that it had closed $45 million Series C funding from Merrill Lynch PCG and other VCs. Matisse's EtherBurst switch is one of the few companies in the Optical Burst Switching (OBS) Domain.
OBS attempts to bridge the gap between optical circuit switching technology (which is widely deployed now) and packet switching which is non-existent in the optical domain because optical components cannot match electronic speeds. OBS establishes dedicated circuits for short durations of time by sending a signal in advance in the control domain which reconfigures the network to form the circuit. Data (which is called an optical burst) is then transmitted over this circuit. The network can be then reconfigured to form a different circuit along which a different optical burst may be sent. While OBS is a neat concept and has existed in academic research since the late nineties, it has not been commercially exploited other than some prototype demonstrations.
I am curious as to how the EtherBurst switches Ethernet 1500 MTU size packets over an OBS network since the granularity of a burst will be much much (1000x) larger than an Ethernet packet. May be they aggregate packets to a destination together and then transfer them. But then latency suffers. It would be great if Matisse came up with some performance characteristics such as throughput delivered, latency, etc.
Posted by Amitabha at 7:40 AM
Thursday, September 06, 2007
Yesterday Cisco launched its first 802.11n-based Ethernet access point (AP), the Aironet 1250. This is based on the 802.11n draft 2 which is currently being standardized and will offer 300 Mbps of wireless access connectivity. Moreover, the access point can be powered by wired Ethernet which avoids having power installations at the deployment point. This product is priced at $1299.
I also read a recent research group report which projected that by 2010 most of wired ethernet in LANs will be replaced by 802.11n technology. I am not so sure. While wireless has the advantage of ubiquity, it still lacks consistant steady performance and security. For example, I see a marked difference between viewing video over my home 802.11g network and with a direct ethernet connection. And the issue is not of bandwidth, 54 Mbps is far more than what my DSL provides.
I am also sure that wired ethernet will be dominant in corporate LANs. Dont think a business can imagine it crucial data floating in the air :)
I have also been having some connectivity issues in my home wireless network, and would be glad if any of you could leave a comment on how to solve it. I live in a densely populated area. There are at least 20 wireless networks around my area. When I switch on my laptop it often has problems connecting to my wireless network. It takes quite a bit of time, sometimes 15-20 minutes before I get connectivity. I suspect it is because of running out of wireless channels in the 801.11g frquency domain. Any thoughts?
Posted by Amitabha at 7:15 AM
Thursday, August 23, 2007
I could find some time to attend the second day today at Hot Interconnects at Stanford. Here are some notes. Although some papers may not be directly relate dto the discussions in this blog, I decided to include them anyways.
A Memory-Balanced Linear Pipeline Architecture for Trie-based IP Lookup on FPGA, USC
The trie data structure is used for IP lookup. Efficient pipelining is possible for IP lookup. In current approaches there exists unbalanced memory distribution across stages. Their solution is to improve pipelining in trie lookups.
Building a RCP (Rate Control Protocol) Test Network, Stanford
They build an experimental network to support RCP using NetFPGA implementation and modifying Linux by adding a shim layer inbetween to support RCP.
ElephantTrap: A low cost device for identifying large flows, Stanford
ElephantTrap identifies and measures big flows in a network using random sampling of packets and algorithms to have select the large flows in the network.
Intel invited talk titled On-Die Interconnect and Other Challenges for Chip-Level Multi-Processing
Chips will become increasingly multicore because performance tapes of with increase in transistors in a single core. However, gain from parallelism is lost in communication overhead. Speaker advocated on-chip high speed interconnect which ties cache, memory, network I/O etc to processors. Also advocated ring based topology for interconnect. He talked briefly about Intel terascale processor with 80 cores.
An Analysis of 10-Gigabit Ethernet Protocol Stacks in Multicore
Environments, Virginia Tech
They consider the interaction of protocols with multicore architectures. They consider TCP/IP and iWarp protocols and the interaction between application, TCP/IP stack and network I/O. They show that if application is scheduled on different core than where interrupt processing is (but in same socket), it leads to best performance as CPU load is distributed across cores and it leads to lesser cache misses too. Did not consider MSI/MSI-X or multithreaded applications.
Assessing the Ability of Computation/Communication Overlap and Communication Progress in Modern Interconnects, Queen's University
They consider interplay between processing and I/O so as to have minimum stalling cycles. They consider various MPI based applications over different interconnects: Infiniband, Myricom and 10 Gig Ethernet
Performance Analysis and Evaluation of Mellanox ConnectX InfiniBand
Architecture with Multi-Core Platforms, Ohio State Univ
They demonstrate how ConnectX (the latest generation of Infiniband technology) performs on multicore architectures.
Memory Management Strategies for Data Serving with RDMA, Ohio Supercomputer Center
They demonstrate that virtual to physical memory translation is a considerable overhead in high-speed networks. They demonstrate how to optimize memory registeration to achieve better performance in RDMA
Reducing the Impact of the Memory Wall for I/O Using Cache Injection, Universiy of New Mexico
Cache Injection is a technique by which data is transferred directly fron device to L2/L3 cache. They show the tradeoff between cache injection and prefetching and propose an algorithm to perform efficient cache injection for network I/O,
Posted by Amitabha at 10:49 PM
Thursday, August 16, 2007
Although I do not follow hot chips, and have never blogged about microprocessors, I would love to talk about Sun's UltraSparc T2 launched last week, the first microprocessor ever to support an integrated network I/O support. Yes, UltraSparc T2 which I would prefer to call by its moniker Niagara 2, supports multithreaded dual 10 Gig Ethernet. From product brochures, this chip seems to be a true microsystem, just connect this to RAM and you get it running.
Although Sun has only mentioned some CPU intensive benchmarks such as SPEC int and SPEC float, may be they come up with some network I/O oriented benchmark numbers as well.
Also Sun is releasing all design documents under the GPL license. Wonder what that means? Can someone take up all the technical documentation and fab a chip independently? How does open sourcing work in the hardware market, does it help in added growth, or is this simply a propganda by Sun to get more interest in this chip, and new buyers thereby.
Sun's CEO Jonathan Schwatrz writes a very interesting blog, one of the best by public company CEO. JOnathan stresses that he thinks about Niagara 2 as a commodity, and Sun MicroElectronics division is encouraged to see the chip independent of Sun systems, and to even competitors of Sun. Interesting and encouraging indeed :)
Posted by Amitabha at 9:34 PM
Saturday, August 04, 2007
With the advent of 10 Gig Ethernet, network I/O has become unbelievably fast. This poses substantial challenges for kernel designers, because the erstwhile thinking that I/O is not CPU intensive does not apply any more. With 10 GigE, you can easily exhaust all the CPU in your box just transmitting or receiving raw packets without doing anything substantial.
This poses an interesting dillema, (i) What is the best network I/O you can perform with your box having 10 GigE, and (ii) Given that an application is running in the background and is the producer/ consumer of network I/O, how do you allocate your limited CPU cycles between network I/O and your application. In my PhD dissertation, we tried to solve the above on a macro scale. We developed a stochastic model to predict the best rate at which network I/O may be performed given the application workload on our system. Our work presented at PfldNet 2007 is availeble here.
With the advent of multicore chipsets such as the Intel quad-core Xeon and AMD quad-core Barcelona besides the Sun Microsystems Niagara chip which has existed for almost two years, there is an added dimension to this problem, how to cooperate the application and network I/O across cores. And with applications becoming increasingly multithreaded, this problem is definitely trickier. Since the interplay between CPU cycles, cache misses, memory latencies etc. becomes even more interesting.
I read a recent paper by authors from Virginia Tech and Argonne National Lab illustrating how a single threaded application can be bound to a core to get higher network I/O throughput. It is a very interesting read, would strongly recommend it.
Posted by Amitabha at 11:25 PM
An excellent presentation made by Dr. Glen Kramer, Chief Scientist at Teknovus at the ITU-T/IEEE Workshop on Carrier-Class Ethernet in Geneva Switzerland is available here.
The presentation motivates the need to upgrade to a 10 Gig EPON standard showing the high bandwidth requirement from a Multiple Dwelling Unit (MDU) with large number of EPON subscribers. Challenges in the standardization activity, particularly meeting the power budget are documented.
Although Dr. Kramer puts some emphasis on the bandwidth requirements of HDTV broadcast TV and 4th generation mobile, I believe that the greatest demand is going to depend on the success of Video on Demand (VoD). I believer that I have aired my views on the potential of VoD before, please read my thoughts here.
Posted by Amitabha at 8:11 PM
Wednesday, July 25, 2007
Quick update, two days ago Intel announced its 10 Gig Ethernet card based on its I/O accelaration technology. Intel is promoting an alternative to TCP and RDMA offload wuth a card that can work better with the processor's I/O capabilities. Basically Intel's strategy is to promote its processors while discounting ethernet cards. This product will a[pparently be launched in September with Inte's Caneland platform serving 4-socket 4-core Xeon processors. I will dig up some more research literature about this product in the coming days.
Posted by Amitabha at 11:52 PM
Friday, July 20, 2007
Quick update, from the archives of the IEEE HSSG public reflector, the PAR and 5 Criteria has been approved by the HSSG. The objectives now include supporting 40 Gig Ethernet over a distance of of 100m, and support 100 Gig Ethernet over a distance of 40km on single-mode fiber. Support will also be provided for OTN.
This is a big win for 40 Gig E proponents, whose support was considerably weak until January this year. It will be interesting to see how the working group handles the tricky issue of dealing with 40 Gig and 100 Gig physical layer challenges simultaneously.
Posted by Amitabha at 12:53 AM
Saturday, July 14, 2007
Thursday, July 12, 2007
Next week at the IEEE 802.3 plenary in SanFrancisco, the High Speed Study Group (HSSG) will meet to gather consensus on a 100 Gbps Ethernet standardization activity. There has been considerable support within the HSSG committee for an interim 40 Gig Ethernet standard, or a standard which is interoperable between 40 Gig and 100 Gig. There have been several news reports covering this battle since the last plenary meet in Geneva in May 2007, unfortinately I have not found time to research this and write about it in the past month in this blog. Here are some interesting articles that I found on a Google search:
Controversy threatens 100G standards work
Group pushes 100 Gigabit Ethernet
To those unfamiliar with the process of standardization, the Study Group has to develop sufficient consensus to have a Project Authorization Request (PAR) and resposes to 5 Criteria which cover (i) Broad Market Potential (ii) Compatibility with existing standards (iii) Distinct Identity (iv) Technical Feasibility and (v) Economic Feasibility. I believe the 5 Criteria of the HSSG has already been decided. Once the Study Group develops consensus, the IEEE forms a standardization committee which develops the standard (also known as the task force). If the HSSG is unable to gather a consensus, then the IEEE may scrap the creation of the standard or create a new Study group with new deadlines. The HSSG has already taken 2 extensions to submitting the PAR.
While I will keep you posted about developments in next week's plenary, please do post your comments on this issue. I am of the opinion that wif a technology is sufficiently mature, it ought to be standardized irrespective of what volumes and market it may drive. Also the standard must taken into account driving a low price point for the technology so that volumes grow up. As an example, although the 10 Gig Ethernet standard is about a couple of years old, 10 Gig NIC volumes are still verry low. One of the reasons I see here is that the standard mandates support for a distance over 300 mts which drives the cost of the PHY devices very high, and thus prevents lowering the entry price point of this technology. Hope the same doesn't happen to the HSSG standard.
Posted by Amitabha at 8:35 AM
Thursday, July 05, 2007
Apologies for not having any blog updates for about a month, it has been a pretty dry June, and I was vacationing for a bit. Am now recharged :), ready to blog for another year. By the way, this September will mark the two year anniversary of this blog.
The Infonetics Research PON and FTTH Equipment and Subscribers" report released last week was very encouraging. Worldwide PON sales are up to $336 million in the first quarter of 2007, while Ethernet FTTH (including EPON) sales are up to $81 million.
Some other good news are:
- Korea is on track to add 700,000 to 800,000 EPON subscribers by the end of 2007.
- EPON subscribers make up 64% of worldwide subscribers, BPON 33%, GPON 3% in 2006. However, this market share will change very soon as Verizon deploys GPON in 2007. The report also quotes Verizon already having 500,000 FiOS customers.
TThe report also predicts high growth for FTTH technologies as service providers aim to deliver more and more bandwidth to the end-user to provide enhanced quality video-on-demand, peer-to-peer networking, online gaming and other high bandiwdth services.
Posted by Amitabha at 9:39 AM
Thursday, May 31, 2007
A product release that escaped my notice: Chelsio released it multi port Gigabit Ethernet adapters last week, its first venture into the lucrative gigabit ethernet market. Chelsio now has the two-port S302E and S302X adapaters and the Quad port S304E adapters. Chelsio mentions in its press release that it will target the storage network and the servers market and will target the replacement of infiniband and fiber-channel technologies with Ethernet.
This is an interesting and ambitious ploy, given the dominance of fiber channel in storage networks. I am not convinced that Gigabit Ethernet is a good technology for these markets in the long run, Chelsio's venture seems to be more to have some product base in this market. Moreover the cards are priced high at $795 and $1495 for the two-port and quad ports respectively. They come with Chelsio's standard TCP/IP offload and RDMA offload technologies, which again I am not convinced is required for Gigabit Ethernet.
I am also quite dissapointed at the lack of ingenuity by Chelsio, two-port and quad-port Ethernet adapters are already available from Sun Microsystems (Sun's product line in this area is available here). Using the same chipset technology but replacing a 10 Gig port by a few 1 Gigs certainly doesn't seem novel enough. Market timing seems to be poor since competitors like Sun already have a much cheaper version of these cards.
Posted by Amitabha at 9:37 AM
Tuesday, May 22, 2007
Taiwan based Hitron and Japan based Sumitomo electric will jointly supply Chungwa Telecom of Taiwan with EPON OLT and ONU products. The deal is expected to total NT$130 million (US$3.9 million). Impressive budget for EPON deployment in Taiwan.
Posted by Amitabha at 7:19 PM
Monday, May 21, 2007
Lightreading reported that Broadcom today announced its eight-port octal physical layer (PHY) device that extends the reach of Ethernet over twisted pair cables. The new 65 nanometer CMOS octal PHY, featuring Broadcom BroadR-Reach will provide added flexibility to standard Ethernet cables by enabling 10/100 Ethernet to operate up to 500 meters on one, two or four wire pairs of Category 5 (CAT 5) or telephony grade cabling, and 1 Gig Ethernet to a distance of 100 metres. This technology will be demonstrated at Interop 2007 in Las Vegas, Nevada.
Posted by Amitabha at 8:05 PM
Yet another manufacturer of Infiniband technologies is now targeting the 10 Gig Ethernet adapter space. Mellanox today announced its 10 Gig Ethernet adapter. It features are not very different from others in the market: it supports multi-core chips, and mentions that it attempts to target the virtualization space. Mellanox also mentioned today that it has shipped over 2 million Infiniband ports. It is interesting to find many infiniband manufacturers releasing ethernet adapters.
Posted by Amitabha at 7:59 PM
Monday, May 14, 2007
Broadcom today released its own 10 Gig Ethernet NIC, supporting Microsoft's Windows TCP chimney engine, iSCSI block storage and remote direct memory access (RDMA) on-chip. It is interesting to see so many players in the 10 Gig Ethernet NIC market now, while the size of the market is still small and uncertain. I will try to come up with a comparison of major 10 Gig Ethernet NIc cards and their features in the next post.
Posted by Amitabha at 10:33 PM
Wednesday, May 09, 2007
Intransa today announced the launch of 10 Gig Ethernet based StorStac Storage System PCU100 SAN. Neterion's 10 Gig adapters will be used by Intransa. Something very relevant to my research work and past discussions in this forum:
Douglas F. Gibbs, PhD, pathology bioinformatics manager at The University of Michigan in Ann Arbor, who tested an Intransa StorStac System 10GbE PCU100 configuration during the beta trials mentioned, “Many of our initial test scenarios revealed that factors other than the Intransa system were rate-limiting in the tests. It will take some catch-up on the part of OS, software and hardware vendors to take full advantage of the 10 GbE speed,” Its great to see more recognition in CPU and OS bottlenecks to deliver the I/O required from 10 Gig E networks.
Posted by Amitabha at 1:14 PM
Tuesday, May 08, 2007
According to a Teknovus press release, FiberHome's FTTH solution using Teknovus' EPON chipsets will be used as a benchmark by China Telecom to evaluate other EPON systems and solutions. FiberHome already has deployments in several provinces in China, Teknovus' aggressive strategy in China seems to be paying off, and more widespread EPON deployments could help drive huge volumes for Teknovus' EPON silicon.
Posted by Amitabha at 6:47 PM
Saturday, May 05, 2007
An excellent paper by Wenji Wu, Matt Crawford and Mark Bowden of Fermi Labs, DOE, titled "The Performance Analysis of Linux Networking – Packet Receiving" is available here.
The authors trace the delivery of a packet from the Network Interface Card to the Application Layer consuming the packet, in the Linux protocol stack. They discuss the operating systems subtelities involved in this entire process and demonstrate the challanges in delivering high network throughput to the application layer. Experimental results discuss the impact of changing some parameters such as the number of packet descriptors to improve network performance. Must read for those interested in high-speed network performance issues.
Posted by Amitabha at 5:32 PM
Thursday, May 03, 2007
LightReading has an article here about the ongoing Ethernet Expo. The article mentions comments about Provider Backbone Transport (PBT) technology including new releases from Extreme Networks. The article also mentions that Hatteras has a new customer -- Mila, the recently spun-off wholesale division of Iceland's incumbent operator, Comments invited from attendees at Ethernet Expo about interesting products.
Posted by Amitabha at 7:24 PM
Tuesday, May 01, 2007
Teak Technologies previewed its pioneering 10 Gig Congestion-Free Ethernet (CFE) switching solution today. The idea mentioned here is to have Ethernet switches collaborate with each other and thereby respond reactively to congestion events rather than perform individually in isolated islands. They mention that the average end-to-end goodput is much better than current 10 Gig Ethernet switches. I am not very sure how this "collaborative technology" works, and no details are available on their website. Comments invited on how this technology works.
Moreover, Neterion announced today that they will partner with Teak Technologies, and provide them with ASIC to deliver an integrated solution to IBM's Blade Centers as part of the Blade.org effort.
Posted by Amitabha at 9:26 PM
Thursday, April 26, 2007
An excellent article available here. The author discusses 10 Gig Ethernet with competitive technologies such as iSCSI and Fiber Channel. The author argues that while the per port cost of 10 Gig Ethernet may be high, administration costs due to the ubiquity and ease of Ethernet technology will be much lower. The author expects costs of 10 Gig adapters to reach commodity price levels in another 3 years. One key area in which 10 Gig Ethernet has room for improvement is end-to-end latency, and reliable delivery of data.
Posted by Amitabha at 10:28 PM
Chelsio today announced the availability of its T3-based Unified Wire Accelerators, which enable server networking, storage networking and clustering on a single platform for the IBM BladeCenter enterprise servers. This card offloads TCP/IP, iSCSI and iWARP RDMA and consumes 16W, which is quite a good number for today's high-speed adapters.
Posted by Amitabha at 10:21 PM
Sunday, April 22, 2007
Woven Systems launched its EFX 1000 Ethernet Fabric Switch with 144 10-Gig Ethernet ports. Woven, a startup founded in 2003, received its Series A VC funding in 2005.
Woven says its primary goal has been to bring 10 Gig Ethernet closer to Infiniband in terms of latency and power consumption. Although Ethernet has the advantages of ubiquity, its power consumption is about 70 watts per port, much higher than Infiniband which is at few watts per port. Similarly, latencies at heavy load may be considerably higher than Infiniband.
The EFX 1000 uses about 16 watts per port and delivers an end-to-end latency of 4 microseconds. The switch also can dynamically load balance data traffic across the various 10 GbE paths. For this, it has its vSCALE packet processing ASIC which uses active congestion management to dynamically monitor congestion across a large fabric and steer traffic onto less congested alternate paths, without dropping or re-ordering packets.
It is evident that Woven Systems is targeting the convergence of wide-area network (WAN) and storage with 10 Gig Ethernet technology. Currently, fiber channel is usually used in Storage Area Networks, while Ethernet is used in WANs. The primary customers are expected to be high-performance computing communities such as national labs and web-server farms.
The product brief is available here.
Posted by Amitabha at 8:15 AM
Thursday, April 19, 2007
In a latest research report, a cumulative total of nearly $64 billion will be spent on metro Ethernet equipment worldwide between 2006 and 2010. Worldwide metro Ethernet equipment manufacturer revenue jumped 83% in 2006, reaching just under $9.2 billion, and is projected to grow 73% to almost $16 billion by 2010.
Posted by Amitabha at 10:52 PM
Friday, April 06, 2007
According to a Infonetics Research report, PON sales grew to $965 million in 2006. The study also mentions that as expected, BPON sales are dropping while EPON and GPON sales are picking up. There is no winner between these two technologies yet, but an overall forecast of a $2.4 Billion PON market looks very promising to the EPON vendors.
Posted by Amitabha at 12:15 AM
Tuesday, April 03, 2007
In a very interesting development, Sun Microsystems today announced that Marvell will be licensing Sun's 10 Gig Ethernet technology to build its own 10 Gig Ethernet cards. More importantly, this technology collaboration is expected to continue to further generations of Ethernet such as 100 Gig Ethernet.
This is a big win for Sun whose multi threaded networking technology employs multiple DMA channels to use multi-core processors that will dominate in the future. This also illustrates how Sun is trying to promote its technology to self-exist rather than limit it to its own products as has happened in the past. Sun's 10 Gig cards were a late entry to the market compared to Chelsio and Neterion's cards.
A detailed press release is available here.
Posted by Amitabha at 9:23 PM
Monday, April 02, 2007
Brocade, one of the powerhouses in Storage Area Network (SAN) switches today announced that it will integrate 10 Gig Ethernet and iSCSI connectivity to its SAN switches. This is in addition to the existing fiber channel technology Brcade uses now.
Posted by Amitabha at 1:47 PM
Sunday, April 01, 2007
Teknovus has had three press releases last week in conjunction with the Optical Fiber Conference (OFC) conference at Anaheim. Teknovus will be using OE Solutions transceivers in its Turbo ONU EPON products. Teknovus will also be teaming up with Cavium Networks and will be using Cavium's OCTEON MIPS64 processor to build a high-performance home gateway to support triple play services. Teknovus also announced that Furukawa will be marketing a Small Form-factor Pluggable(SFP) ONU using its TK3713 chip.
Any reader who visited Teknovus's booth at OFC, it will be great to have your comments on their demos.
Posted by Amitabha at 11:12 PM
Tuesday, March 13, 2007
In an interesting post on LightReading today, AT&T is said to be favoring Etehrnet over copper using VDSL technologies for last mile solutions. While VDSL integrated with EPON has been discussed before, the idea of having Ethernet over DSL is interesting. The article also mentions the absence of startups in this area, with only one startup Aktino having related technology. It is also interesting to observe this thought when PON deployments in the US are moving towards the GFP-PON world. Here is the link to the article.
Posted by Amitabha at 6:59 PM
Tuesday, March 06, 2007
Tuesday, February 27, 2007
End-to-end Ethernet is big now, and Ciena is the latest to market solutions for end-to-end Ethernet. Yesterday, Ciena inroduced the CN 5060 Multiservice Carrier Ethernet Platform, which it says is currently in trials with service providers. The CN 5060 delivers E-Line and E-LAN services, advanced Ethernet interworking for Layer 2 services, MPLS functionality, and support for TDM-over-packet (TDMoP) capabilities to enable service providers to deliver any service over a performance-grade Carrier Ethernet network.
Posted by Amitabha at 8:33 AM
Immenstar, the EPON chipset vendor founded in 2004 was acquired yesterday by Cortina Systmes. The terms of the deal are undisclosed. The press release mentions that Cortina aims to achieve an end-to-end solution, and the Immenstar EPON solutions fit into its plans for the access world.
Immenstar getting acquired is not a big surprise for me, because the company always looked like one of those which will deliver some hot stuff, and then make a quick exit. It operated on stealth mode for quite a while, and emerged suddenly in 2004 with EPON solutions.
So now the bug question is: What are the plans of Teknovus? It is probably the only startup left with an established product and customer line, and therefore must be a big acquisition target for any company aiming to enter this space. And an acquisition probably about to happen soon, because it doesnt seem that the EPON market is growing fast enough for the investors at Teknovus to withstand all the competitive pressures from large companies.
Posted by Amitabha at 8:05 AM
Wednesday, February 21, 2007
The official specs of the Sun 10 Gig Ethernet Adapter are available here. Its interesting to see Sun price the Ethernet card and transceiver separately. This card doesn't mention the multi-threaded CPU support, so probably there will be further versions of this adapter that support this feature.
Posted by Amitabha at 10:06 PM
Tuesday, February 20, 2007
In an interesting and remarkable development, Sun unveiled its multi-threaded 10 Gig Ethernet technologies. The idea is to employ multiple DMA channels (as many as 40) in parallel to ensure simultaneous data transfer to be handled by multiple processors in a multi-core architecture. Sun mentioned that this solution yields data transfer 4 times faster than using existing Ethernet Adapters (which I believe refers to Neterion's technology, since thats what Sun has been mainly supporting)
This demonstrates how serious 10 Gig Ethernet is going to be, particularly for end-to-end data transfer to high-end servers. Although Sun may not have an immediate market, I believe that its serious adoption of 10 Gig technology will help it emerge as a leader in this pace.
The 10 GigE effort discussed earlier with the codename 'Neptune' in this forum is expected to be part of every major Sun technology including the Niagara 2 and Rock servers expected to be released later this year.
Posted by Amitabha at 11:15 PM
Wednesday, February 14, 2007
Centillium recently released its EPON ONU Mustang 200 ONU System on Chip (SOC). This ONU may be operated with Centillium's Colt OLT, or any other 802.3ah compliant OLT. The produce specifications are available here. In brief, the Mustang can support 8 classes of service and 256 multicast addresses.
Posted by Amitabha at 8:24 AM
Thursday, February 08, 2007
Robert Metcalfe, the inventor of Ethernet, is one of the 18 inventors to be inducate d into the National Inventors Hall of Fame in the United States. Other people chosen this year include the inventor of MRI. Congratulations!!!
Posted by Amitabha at 9:53 AM
Wednesday, February 07, 2007
LightReading has an interesting article here on Meriton's plans for a new technology called Carrier Ethernet Transport (CET). The idea is to switch Gigabit Ethernet tunnels directly on the WDM layer instead of doing it a layer 3. The article talks about end-to-end Ethernet delivered via backbone networks and the Ethernet PON. If anyone has some more insights into this technology please post a comment on this forum.
Posted by Amitabha at 6:12 AM
Wednesday, January 31, 2007
Cisco today announced that it will be providing 10 Gig Ethernet and Power over Ethernet support in its Catalyst switches which are deployed in enterprise/ metro area networks. One of the prime goals is to support high definition video in enterprises. Also targeted is IP-telephony, so that telephones can draw all the required power supply from the network.
Posted by Amitabha at 11:59 PM
Sunday, January 28, 2007
This just came to my attention, the IEEE 802.3 has formed a study group to investigate on Energy Efficient Ethernet (EEE). This study group met for the first time at the IEEE meets at Monterey. This effort is being led mainly by scientists from the Lawrence Brekeley National Laboratory, and includes participation from Cisco and Broadcom.
The outstanding issues document available here is an interesting read. It mentions that EEE will save 250 to 380 million $s a year from 1 Gig Ethernet and 40 to 80 million $s a year from 10 Gig Etehrnet. The main idea here is that Ethernet cards should be able to switch to lower data rates when link utilization is low. As an example, 1 Gig cards should be able to switch quickly between rates of 10 Mbps, 100 Mpbs, and 1 Gbps depending on link utilization. An interesting question is whether upper layers need to be made aware of the switching of rates?
Posted by Amitabha at 8:15 PM
Wednesday, January 24, 2007
This is an interesting article at Lightreading reporting cable companies facing a bandwidth crunch. This is not unexpected, since residential broadband bandwidth consumption is increasing rapidly, particularly downloaded video traffic, such as from YouTube. Moreover, the data channels available in DOCSIS 2.0 are limited, DOCSIS 3.0 will still take some time to deploy.
What I found more interesting is the mention of PON under consideration by cable companies. Cable deployments already have fiber to a head end from where a copper ring takes over. Unfortunately the copper ring is shared by a lot of customers, may be even 1000. This is because cable was designed for broadcast, and a lot of customers could feed from the pipe with little attenuation to the signal. However, this architecture is not convenient for data traffic which is not shared between customers. Similarly, upstream traffic performance is pretty poor as the number of upstream channels in DOCSIS 2.0 are very less.
PON will definitely help a long way. I am not sure why PONS have been completely dismissed by cable companies till now. With the competition from satellite television which delivers a much wider range of HDTV channels, and the phone companies BPON architecture, rapid developments in cable may be expected soon.
Posted by Amitabha at 8:09 AM
Tuesday, January 23, 2007
Teknovus yesterday announced that it would use Fiberxon's dual-speed transceiver's in its Turbo EPON (2.5 Gbps) products. The Turb EPON can support the legacy bit rate of 1.25 Gbps as well as the turbo mode of 2.5 Gbps. This is Teknovus's solution to match the speed of the GPON while the 10 Gig EPON standardization is going on. The press release is available here.
Posted by Amitabha at 10:13 PM
Monday, January 22, 2007
Sunday, January 21, 2007
Thursday, January 18, 2007
Well this news is a surprise to me. Sun is coming up with its own 10 Gig Ethernet chip, which is called Neptune. Apparently, this is the first time that Sun is coming up with its own ASIC interface card. Neptune will be optimized for multithreaded applications and will be designed to work well with multicore chipsets.
Sun has already been working with Neterion and using Neterion's adapters in its Sun Fire servers, but I guess the performance has not been upto Sun's standards, which is making Sun come up with its own chipsets. According to this news report Sun believes it can achieve 4 times better performance with its own ASIC for the Niagara chipset. This report also mentions Neptune will intelligently allocate threads to the correct cores on the eight core, 32-threaded Niagara processor and will also head-of-line blocking problems. Neptune will support dual and quad-port Ethernet connections running at 10 or 1Gbit/second in a single PCI Express card.
Sun is also coming up with its next generation processor to Niagara called "Rock" which will be released sometime in 2008. With the standardization efforts of 100 Gig Ethernet, I believe Sun wants to enter the adapter market, which is very important for it to deliver high-performance servers.
Posted by Amitabha at 7:52 AM
Tuesday, January 16, 2007
Well Nortel's PBT kept making major news today as well, with the deal with BT sending Nortel's stock up 3.5% today. And Philippe Morin, president of Nortel's Metro Ethernet Networks division mentioned that he expected Nortel to have a major first to market advantage with PBT, and to make a major dent in the carrier ethernet market in the next year. Currently Nortel has only 4% market share in Carrier Ethernet, with Cisco leading at 57%. It will be pretty interesting to see how PBT moves along.
Posted by Amitabha at 8:19 PM
Monday, January 15, 2007
In a widely expected move, British Telecom signed deals with Nortel and Siemens for Provider Backbone Transport (PBT) based metro ethernet switches. PBT is a very new and unstandardized option over MPLS for carrier ethernet.
BT is the first to support this technology and it is expected that companies like Verizon will show interest soon. This deal may be a big turnaround for Nortel as it was one of the pioneers for PBT and competing companies like Cisco and Juniper are yet to enter the fray.
PBT is a technology to provide an end-to-end virtual connection over a network, so that Quality of Service (QoS) features and Service Level Agreements (SLAs) may be made available. A great insight into this technology is available here.
Posted by Amitabha at 9:09 AM
Friday, January 12, 2007
Lightreading has an interesting article about the competition between Teknovus and PMC-Sierra to capture segments of the Asian broadband access market. It mentions Teknovus' CEO Rex Naden mentioning that Teknovus still is hopeful of getting a big contract in Korea as KT moves towards increased deployments and possibly migrate to higher bit-rates (Teknovus supports non standardized EPON solution at 2.5 Gbps).
Competition seems intense between these two leading EPON vendors, because volume deployments of EPON have still not caught up except for in 2 countries, Korea and Japan, which have a combined population of 175 million. On the other hand , EPON is now a very mature industry, and price per port is spiraling down. Since the customers are vendors who sell to a consumer market which is not willing to afford more than $100 per box (and mostly this cost is absorbed by service provider), the margin per port is very low. It is a difficult business unless deployment picks up to tens of million ports per year.
In this light, I believe that its best for PMC-Sierra and Teknovus to join hands and target GPON competition. May be its too late, since North America is heavily inclined towards GPON at this point of time. But there are the big consumer markets - India and China. And Europe is an emerging market too, though many cities such as Paris and Amsterdam have preferred active ethernet.
But often, when two leading competitors collaborate, they increase the market size, which is mutually beneficial. Take the case of Google and AskJeeves in early 2000-2001. Collaboration helped them expand search, and see the size of the search market now.
So my suggestion to both companies is: collaborate against competing technologies. Broadband access market is huge. With positive developments both companies may have billion dollar revenues by the turn of the decade.
Posted by Amitabha at 4:52 AM
Thursday, January 11, 2007
Just read a news article about Force 10's focus on patenting high-speed backplane technology. For example, US Patent 7124502 titled "Method of patenting a high-layer-count backplane" has been recently patented by Force 10. I have to read teh patent to find out the speed this can support. Force 10 already has its Terascale E-Series routers which can support a backplane bandwidth of 5 Tbps, and is currently used for 10 Gig Ethernet switches.
Posted by Amitabha at 8:58 PM
Tuesday, January 09, 2007
Saturday, January 06, 2007
I read this article about a Cisco Engineer commuting 186 miles one way from Mariposa to San Jose to work, 5 days a week. This guy won the prize for America's longest work commute and earned a prize money of $10,000 besides other things. I am amazed that there is a competition of this kind. And more amazed that the winner is an employee of a company whose mission states that the "Internet will change the way we work, live, play, and learn".
The key to achieve this is high-speed broadband access, and I would like Cisco and competing companies to invest more money in this. Yes, the Internet is great, it has revolutionized and shrunk the world, communications have become so cheap, but the ultimate test is delivering unprecedented and cheap bandwidth to all users. So that we may communicate in better and more media-rich ways, and distances can be further bridged, and this Cisco Engineer can work from his ranch in Mariposa. Think of how much energy we all could save if we could reduce the work commute, and allow everyone to work from the comfortable environs of home.
So my hope for 2007 is that high-speed broadband access technologies like EPON can start off in the United States. Korea and Japan have leapfrogged ahead in this technology, countries in Europe have also taken the same up.
Posted by Amitabha at 4:08 PM
Friday, January 05, 2007
Happy new year to all readers.
The big news reported on Lightreading today is about Korea Telecom (KT) going ahead with the deployment of 800,000 EPON connections. This is the first large-scale deployment of EPON in Korea. Although KT has been investigating PON including WDM-PON for quite some time now, the target of close to a million subscribers in 2007 is big news.
The main equipment builder is expected to be Dasan which will be using PMC-Sierra (Passave) chipsets.
Posted by Amitabha at 6:40 PM