Skip navigation

Future Intel® Ethernet Server LOMs (LAN On Motherboard) will have a new technology that will allow the server to communicate with the on-board BMC (Baseboard Management Controller) over the LOM via standard Ethernet.  The technology is called OS2BMC.




Why is this important? Consider a datacenter where you have hundreds or even thousands of servers, all with BMCs for monitoring the health of the individual servers.  There is usually some sort of management application running in the datacenter to actively monitor those systems remotely over the network.  The management application sends packets over the Ethernet to a server, where the Intel® Ethernet Server LOM examines the packet and determines the packet is destined for the BMC (instead of the host) and sends it to the BMC.




This is a very reliable, cost effective and well-known management paradigm.  Now consider that you have several of these servers dedicated to monitoring and managing the other servers in the datacenter.  These servers can communicate with all the other BMCs in the datacenter via the network, however they cannot communicate with the local BMC over the Ethernet – it must use a local interface that is completely different than Ethernet.




This essentially requires the management application to be able to communicate via two different mechanisms (Ethernet and local), resulting in additional engineering and evaluation tasks.




The OS2BMC technology in upcoming Intel® Ethernet Server LOMs provides a communication path to the local BMC over the Ethernet, simplifying the management application significantly.




I’ve created a video for your viewing that provides a good background for why this technology is cool, and how it works.



Please take a watch and let me know what you think.




We on the Intel® Ethernet Virtualization team have just finished up the 2nd paper in a series about Ethernet in a virtualized environment.  The 1st paper is a study in best practices with VMware® ESX 4.0 and 10Gb Ethernet, as introduced in an earlier blog:



The new paper discusses how one might need to re-imagine the way the Ethernet in the datacenter looks when you introduce racks of servers employing virtualization solutions.  Historically a rack of servers had two or more 1Gb Ethernet uplinks to a top of rack switch, which in turn was uplinked usually to a 10Gb Ethernet backbone in the datacenter.



With virtualization, you can imagine that the virtual machines have Ethernet uplinks to the virtual switch within the Hypervisor/VMM – this becomes essentially what was traditionally a top of rack switch, connecting all the virtual machines in the physical server within a rack.  Now all of these physical servers employing virtual switches can be uplinked with 10Gb Ethernet connection to a 10Gb Ethernet switch at the top of the rack, which in turn is still connected to the 10Gb Ethernet backbone in the datacenter.




This latest paper is available here:


We had printouts of this at IDF just last week and it was well received, with many well-known companies looking to re-brand the document and publish it themselves.




Give it a read and let us know what you think.


One Year of Wired Ethernet

Posted by dougb Sep 17, 2010

Happy Birthday!

September 17, 2009, we launched this community to help bring together all of the millions of wired Intel® Ethernet users and developers across the world in one place.    The launch was somewhat of a “build it and they will come” type of effort, and now, as we head into our sophomore year, it’s time to reflect on what we are and what we should become in what, here in the Farm, we like to call “Stop, Start, and Continue.”  Since this is supposed to be a conversation, please feel free to jump into the comments and provide your inputs.  Here are my thoughts.

1)      Continue – Providing how-tos, tips, tricks and ways to leverage your investment in Intel® Ethernet.  We need ideas of the types of things you need help with, so please let us know.

2)      Start – Some of you are struggling with issues with our products.  First, we are listening – we hear you.  Second, we are feeding a lot of your concerns into our product teams.  We know we need to tell you more about what we are doing in the threads.  There are rules about what we can say and what we can’t so please be patient, but we want to give you the information you’re looking for.

3)      Start – The web is a multiple media buffet, and we’d like to offer more video and podcasts here on various topics.  So every once in a while instead of yet another blog posting, we’ll mix in videos and podcasts.   If there is a topic or Intel expert you want to hear from, please let me know.  Online chats are also something we are looking forward to adding in this column.

4)      Stop – There is a saying about web communities that goes something like this:  “If you sell pens, you don’t make a community about pens, you make a community about writing.”  Our community has been about Ethernet.  We’ll provide more value to you by making it more about networking, communications, and solutions built on Ethernet. 

These changes will take a while and will come out slowly over time, but be sure to let me know where we need to provide more value to your experience.

Now come the action items for you out there.

1)      Continue – While it might seem like we aren’t replying to all the threads, we do have somebody that reads them all and brings them to the attention of the right part of the org.  The link issues with the 82567 was brought before the product team and resulted in a modification to the release schedule to move up some improvements.  Only by posting will that type of influence continue.  So please let us know what we are doing right and what we could be doing better in relation to our product usage.

2)      Start – Get your voice heard!  If you want to see us make a how-to, the only way is to let us know.  Vote in the polls.  Comment on the blog posts that you find interesting.

3)      Stop – Remember, if you have access to Intel premier support, that is the best way to get your support issue handled.  Posting here, SourceForge, mailing lists and every where at once won’t necessarily improve things.  We see all of them and answer shopping is not the way.  We have business processes that dictate a lot of what and when, which makes it the why if you’re not getting an immediate answer on any one of the support vectors.  We want to keep earning your business, but we don’t have a ton of people and priorities which sometimes mean difficult ordering to the solutions.  I hope you understand, and we’ll try to do better.

Now for some random ideas:  If we held an Intel® Ethernet Networking summit, would you attend?  If we held online classes would you attend?  If either of these are something you would do, just post a reply and we’ll do more thinking about it.  If you know of a tradeshow where we should visit and do a class, please let us know.

I hope this year of progress has helped your experience with Wired Intel® Ethernet be more positive.  I look forward to continuing to help this community.


SR-IOV Explained

Posted by dougb Sep 7, 2010

Virtual Machine Device Queues (VMDQ) is a great technology, and our virtualization master, Patrick, explained that in part one.  In this segment, Patrick talks about the next revolution in Virtualization, Single Root I/O Virtualization.  Again Patrick doesn’t drop you into the deep end, but eases into it before showing the power of this advanced feature.

Post in the comments section the types of things you want Patrick to think about for his next how-to segment on Virtualization.

Filter Blog

By date: By tag: