Cisco Logo


Data Center and Cloud

In a recent post, I read that there’s no reason to wait to implement FCoE and that an underlying lossless Ethernet network is not really required. In fact, a CEO of another vendor actually proclaimed that their switch is lossless and can be used for FCoE today.In some ways, I agree. There is no reason to wait to implement FCoE today. But shouldn’t you implement it with all the features necessary for a truly lossless Ethernet? Features like Priority Flow Control (802.1Qbb), Bandwidth Management (802.1Qaz), and Congestion Management (802.1Qau)? All of these encompass what we call Data Center Ethernet and are being standardized by IEEE and supported by the Ethernet Alliance.It’s also interesting that he measures lossless capability by testing just simple forwarding on a single switch. Any modern switch can forward line-rate at different packet sizes with zero packet loss. The real question is how does an Ethernet switch handle congestion? How then does it handle storage traffic that doesn’t tolerate drops while dealing with congestion of other protocols? This is why Data Center Ethernet is required which allows you to build a lossless NETWORK, not just a lossless switch.The Cisco Nexus 5000 delivers Data Center Ethernet and FCoE today. Be sure to look at it when comparing against other so-called lossless Ethernet switches.

In an effort to keep conversations fresh, Cisco Blogs closes comments after 90 days. Please visit the Cisco Blogs hub page for the latest content.

2 Comments.


  1. Dante Malagrino
    Dante Malagrino

    Very well said, Deepak. The truth is that FCoE is real and all of Cisco’s competitors rushing to get something (even just PPT decks) out is a proof that customer demands is there.If FCoE had not gotten so much traction, we would not be reading so much about it and we would not be arguing so much on who does it better ;-)– Dante

       0 likes

  2. I agree with what is being said. I think everyone is only looking at FCOE though. Lossless provides several advantages to ISCSI and NFS. The virualization world (not to mention database clustering/RAC) is using both as ISAN/NAS access protocols for backend datacenter ethernet networks. I am not sure you can justify FCOE at all unless you are maintaining your current server fc connections with a new consolidated network. The mistake I see being made is designing net new for FCOE. It should be 10g lossless iscsi for block level or 10g nfs/cifs for file level. There is no reason to introduce the complexities of FC when an all ethernet network will possible even exceed the throughput. Imagine not having to maintain that wwn port map excel sheet…

       1 like

  1. Return to Countries/Regions
  2. Return to Home
  1. All Data Center and Cloud
  2. Return to Home