Cisco readying a new Nexus switch

2013-01-28 11:15:10

Week News Abstract For Fiber Series in 10GTEK
The abstract is mainly about the optical communication related products,including: SFP,QSFP,FTTH,GPON,EPON,SFPPLC,PTN,ODN,Sfp Transceiver,Optic Transceiver,Optical module,Optical devices,optical communications,Optical transceiver module,Etc.

Brocade guns for 100G Ethernet supremacy
The advancements, announced at the Brocade Technology Day here, are intended to support cloud environments with thousands of virtual machines running within and between data centers. The chassis switch complies with Brocade's VCS Ethernet fabric, which was developed to provide a lossless, low latency, deterministic multi-path Ethernet network based on the TRILL specification in the forwarding plane.The new wares will undoubtedly heat up the fabric switching competition between Brocade and Cisco, HP and Juniper. But none of these players yet has introduced a high-density 100G Ethernet switches for the data center core -- Cisco's play is believed to be startup and potential spin-in Insieme Networks. And to date, Huawei is the only major vendor to have shown a high-density 100G Ethernet switch.Brocade, the leader in FibreChannel storage-area network switching, is looking to maintain its hold on the data center as competitors -- and Brocade as well -- offer converged Ethernet LAN, storage and compute systems. In its effort on this front, Brocade acquired Ethernet switch vendor Foundry Networks in 2008 but maintains a less than 2% share of the overall market.Perhaps Brocade's new switches will up that share. The new VDX 8770 chassis switch (pictured above) tops off the VCS-capable VDX switching line from Brocade. It includes 8-slot and 4-slot chassis supporting 4Tbps per slot, Brocade says. The switch was two years in development and was initially expected to ship last year.At 4Tbps, the 8770 could theoretically support 40 100G Ethernet ports per slot (if physical real estate on a module will allow), or perhaps more than 300 per 8-slot chassis. Brocade would not disclose the 100G densities it is aiming for but said 100G modules for the 8770 will be rolled out in the next 12-18 months.That capacity could also conceivably support 400 10G and 100 40G ports per slot. But at first ship, the VDX 8770 will support 48-port Gigabit Ethernet, 48-port 10G and 12-port 40G line cards, all at line rate, Brocade says.The VDX 8770 also features 3.5 microsecond port-to-port latency, and can tunnel VXLAN and NVGRE network virtualization traffic at line rate, the company says. In addition to TRILL support for Layer 2 multipathing, the VDX 8770 also supports Brocade Trunking for Layer 1 resiliency and a technique for Layer 3 multipathing as well.Each 8770 switch also supports 384,000 MAC addresses and VMs, and provides VXLAN and NVGRE tunnel monitoring for visibility of VMs in software-defined networks, Brocade says.The company says it has more than 700 VDX switch customers, and that VDX/VCS fabrics can support from 12 to 8,000 ports. The VDX 8770 will ship in the fourth quarter with prices starting at $65,000, and $833 per 10G port. The list price for 40G is $5,000 per port, not including optics.
Dell: SDN won't turn enterprise switches into commodity gear for 3-5 years
The original promise of SDN (software-defined networking), which puts most higher-level network functions into software, was to make switches into mere forwarding engines that may be interchangeable and cost far less than they do now, said Joshipura, head of product management and marketing at Dell Networking, at a press briefing on Thursday."Three years ago, the big driver of SDN was, 'If I do this, I'll get $1,000 switches,'" Joshipura said. However, cheaper gear isn't the current payoff for those deploying SDN, because most implementations will remain in hybrid networks with traditional gear for a long time, he said."For the next three to five years, until we get to mainstream SDN, cost is not the primary driver of SDN," Joshipura said.Today's enterprise switches need to make many decisions on their own, which requires they be equipped with sophisticated, expensive ASICs (application-specific integrated circuits). SDN is designed to eventually put all network decision-making in controllers, which can be implemented in software and run on standard x86 servers. Easier data-center management and new network capabilities are also seen as eventual benefits of SDN.However, most organizations will migrate to SDN over a long period while holding on to their investments in traditional gear, Joshipura believes. In fact, SDN has barely cracked live deployment so far, being implemented primarily in higher education and in Web companies, he said. Mainstream enterprise adoption will require more enhancements and standardization. "Slowly logic moves out from the switch into a controller," Joshipura said.IT shops will see benefits along the way, Joshipura said. Most importantly, controllers will make it be easier to provision switches, he said.Though Dell originally built its business on standardized, price-driven products, namely PCs, it now may have something to fear from commoditization. More than half of the company's business now is based on its own intellectual property, according to Joshipura. Through a series of acquisitions, including its buyout of high-end networking vendor Force10 last year, Dell is now going up against broad-based giants such as Cisco Systems, Hewlett-Packard and IBM to supply all the elements of a data center.When switches do become cheap commodities, Dell has a plan for remaining relevant in that new world of networking, Joshipura said. It wants to extend SDN beyond its current realm to give organizations more flexibility and management capabilities.Today, through the emerging OpenFlow protocol, SDN has only been standardized between a controller and the so-called data plane, where switches forward packets, he said. Dell wants to offer APIs that reach all the way up to the hypervisor. It plans to optimize its software for the key hypervisors, including VMware's ESX and Microsoft's Hyper-V as well as the open-source OpenStack framework, while remaining hypervisor-agnostic, Joshipura said.
Cisco readying a new Nexus switch
Sources, however, say the Nexus 3500 will feature 250 nanosecond port-to-port latency and integrated network address translation. They say it will give InfiniBand a run for the money, especially when combined with a new NIC from Cisco called usNIC."If this product does come out with latency as stated, it will dominate the silicon industry slamming down on Broadcom and more importantly Fulcrum," the source says. "But also compete with Mellanox and close the gap with InfiniBand. Cisco will be untouchable in ultra low latency switching."RedHat has apparently run tests of usNIC with its MRG-M high performance computing software. Slides 16-18 of this presentation appear to show performance improvements of MRG-M when running usNIC vs. InfiniBand.Red Hat was not immediately available for comment. But our source said it, along with the Nexus 3500, could be a viable alternative to InfiniBand."That, combined with this new Nexus 3500 having 250ns latency would be a compelling solution against InfiniBand," the source said. "If Cisco launches Nexus 3500 in the next few months and combines usNIC in the launch it will finally be the first Ethernet solution that can compete against [InfiniBand] ... [and] shows Cisco intent to kill InfiniBand with Ethernet." Fat chance, says Mellanox."First, it's great to see that Cisco is working to build a solution to compete with InfiniBand," says Mellanox Vice President of Market Development Gilad Shainer. "We see this as yet another testament to the increased InfiniBand market share over Ethernet and Cisco solutions."Ethernet does not provide better performance, scalability and or cost compared to InfiniBand. In particular to Cisco, the InfiniBand cost advantage is even bigger. Furthermore, the use of InfiniHost III adapter (on slide 17 of the deck you've linked to below) was released in 2005 and is not representative of current InfiniBand performance."I'd also like to note that latency is not the entire story here," Shainer continues. "The value of InfiniBand is also in its bandwidth, RDMA, lowest CPU overhead and the unlimited scalability. Ethernet, nor Cisco, can build an infrastructure that is as cost-effective and scalable. This is why we're seeing many large data centers moving to InfiniBand."Cisco currently offers the Nexus 3000 line for low-latency, top-of-rack switching targeted at high-frequency financial trading, as well as the Nexus 7000, 5000 and 2000 lines for data center fabric switching.
The above information is edited by 10GTEK.
10GTEK TRANSCEIVERS CO., LTD (Hereinafter refered to as 10GTEK) is specialized in developing and manufacturing Fiber Optical Transceivers and High Performance Cables which are wildly applied in Datacom, Telecom and CATV, providing customers with top quality and cost effective products. Our High Speed Cables cover Passive SFP+ Cable, Active SFP+ Cable, QSFP+ cables, MiniSAS (SFF-8088) Cables, CX4 Cables, Harness cables, Breakout Cables, Patchcords. We also manufacture Fiber Optic Transceivers like 10G XFP, 10G SFP+, SFP DWDM/ CWDM, GBIC, etc. The prompt response and excellent customer support contribute to clients‘ full satisfaction.Today, 10GTEK has been growing fast in the optical field for its unique and competitve excellence which has got a high attention from datacom and telecom.
?This article reader also like:China to require FTTH for new homes