Comments on: Is the Cisco Nexus 7000 Needed Today – Or Tomorrow ? http://etherealmind.com/cisco-nexus-7000-today-tomorrow/ Network design, architecture, thinking, working. Tech. Fri, 10 Feb 2012 18:43:00 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Venkat http://etherealmind.com/cisco-nexus-7000-today-tomorrow/#comment-155 Venkat Thu, 19 May 2011 07:10:07 +0000 http://etherealmind.com/?p=96#comment-155 Hello Greg / Omar, Now we are in middle of 2011, Whats your thoughts about Nexus 7k and FCoE. How do you see the future from here considering DC consolidation,Virtualisation and Storage. Regards/Venkat Hello Greg / Omar,
Now we are in middle of 2011, Whats your thoughts about Nexus 7k and FCoE. How do you see the future from here considering DC consolidation,Virtualisation and Storage.

Regards/Venkat

]]> By: Omar Sultan http://etherealmind.com/cisco-nexus-7000-today-tomorrow/#comment-154 Omar Sultan Tue, 26 Feb 2008 02:38:59 +0000 http://etherealmind.com/?p=96#comment-154 Greg: Nicely thought, well-written analysis. If you don't mind (and since you have a blog, I assume you don't), I thought I'd add a couple of comments into the mix. :) NEW OS, NEW TECHNOLOGY So, NX-OS is built atop SAN-OS, so we are not starting from scratch, with code that's never been run in a production environment before. That foundation and the modular nature of NX-OS are two keys to solid, stable code. You are correct that we have maintained the management interfaces, so, for the Catalyst crowd, there is an IOS-like CLI that maintains the existing command structure, syntax etc so there is steep learning curve. For the MDS crowd, when unified fabric modules are available, they can manage storage services with what is essentially Fabric Manager. That being said, you are correct that this platform does have different internals, so folks will have to get used to the operational norms for the new platforms FIBRE CHANNEL OVER ETHERNET So, we have built a platform that will serve the needs of FCoE--lossless fabric, storage friendly OS--we have all the pieces needed. We know what the requirements are because we are fully engaged with the standards bodies to get the standard baked. But, you will also notice, we do not have any FCoE modules because, as you point out--there is no standard yet. Its important to note, when unified fabric modules are available, you will not have to upgrade OS, fabric or supervisors to take advantage of them. FCoE does change the dynamics of the data center quite a bit. While some portray this as an assault on Fibre Channel, it really is not--the FC SAN is not going away any time soon. One of the things FCoE will do is give 100% of the initiators access to 100% of the targets, so it will actually enhance the role of the SAN in the data center. NETWORKING VS STORAGE Also a good point. Personally, I believe the organizational transformation challenges may be bigger than the technical challenges. We are still committed to the Catalyst and MDS, so our belief is that people will move to a unified fabric when it makes sense for them--we have no need to push. We are also working on supporting customers through the organizational aspects of unified fabric along with the technical aspects. As far as storage folks specifically, as I noted earlier, a unified fabric actually enhances their value-add. Beyond that, much like the transition from TDM phones to VOIP, voice teams did not go away, we just simplified the transport. FCoE, or iSCSI for that matter, is similar--FC is not going away and we still need storage services and storage architecture, etc, we are simply consolidating the transport. PENETRATING THE SERVER FARMS I think the trick here is to have something interesting to say to the server people--something we have not always done or done well. I think, when the time comes, we will have some compelling messages for them. EARLY UPGRADES I think you will be pleasantly surprised on this front. For example, we have loosely coupled the data plane from the control plane, so taking advantage of next-gen fabric modules does not imply a supervisor upgrade or I/O module upgrade--we call this forward investment protection. MISSING FEATURES The Nexus is not meant to be a universal replacement for all switches. Our belief is that you need to take a portfolio approach to find the correct switch for the particular needs of piece of your data center , so as cool as the Cisco Nexus 7000 is, you will still need rack switches, integrated blade switches, GE capability, 10GbE capability, etc. I actually believe this is a strong point for Cisco: there is a great amount of design flexibility across the Nexus 7000, Catalyst 6500, Catalyst 49xx, Catalyst Blade Switch 31xx/30xx, Cisco 7600, etc while still maintaining consistent management, operations, and features. Also bear in mind that the Nexus is a new family--the Cisco Nexus 7000 is the first in the series, but there will be other form factors focused on other parts of the data center network. Well, that's about it--hopefully this adds to the discussion. Feel free to ping me if you have any questions. Omar Greg:

Nicely thought, well-written analysis. If you don’t mind (and since you have a blog, I assume you don’t), I thought I’d add a couple of comments into the mix. :)

NEW OS, NEW TECHNOLOGY
So, NX-OS is built atop SAN-OS, so we are not starting from scratch, with code that’s never been run in a production environment before. That foundation and the modular nature of NX-OS are two keys to solid, stable code.

You are correct that we have maintained the management interfaces, so, for the Catalyst crowd, there is an IOS-like CLI that maintains the existing command structure, syntax etc so there is steep learning curve. For the MDS crowd, when unified fabric modules are available, they can manage storage services with what is essentially Fabric Manager. That being said, you are correct that this platform does have different internals, so folks will have to get used to the operational norms for the new platforms

FIBRE CHANNEL OVER ETHERNET
So, we have built a platform that will serve the needs of FCoE–lossless fabric, storage friendly OS–we have all the pieces needed. We know what the requirements are because we are fully engaged with the standards bodies to get the standard baked. But, you will also notice, we do not have any FCoE modules because, as you point out–there is no standard yet. Its important to note, when unified fabric modules are available, you will not have to upgrade OS, fabric or supervisors to take advantage of them.

FCoE does change the dynamics of the data center quite a bit. While some portray this as an assault on Fibre Channel, it really is not–the FC SAN is not going away any time soon. One of the things FCoE will do is give 100% of the initiators access to 100% of the targets, so it will actually enhance the role of the SAN in the data center.

NETWORKING VS STORAGE
Also a good point. Personally, I believe the organizational transformation challenges may be bigger than the technical challenges. We are still committed to the Catalyst and MDS, so our belief is that people will move to a unified fabric when it makes sense for them–we have no need to push. We are also working on supporting customers through the organizational aspects of unified fabric along with the technical aspects.

As far as storage folks specifically, as I noted earlier, a unified fabric actually enhances their value-add. Beyond that, much like the transition from TDM phones to VOIP, voice teams did not go away, we just simplified the transport. FCoE, or iSCSI for that matter, is similar–FC is not going away and we still need storage services and storage architecture, etc, we are simply consolidating the transport.

PENETRATING THE SERVER FARMS
I think the trick here is to have something interesting to say to the server people–something we have not always done or done well. I think, when the time comes, we will have some compelling messages for them.

EARLY UPGRADES
I think you will be pleasantly surprised on this front. For example, we have loosely coupled the data plane from the control plane, so taking advantage of next-gen fabric modules does not imply a supervisor upgrade or I/O module upgrade–we call this forward investment protection.

MISSING FEATURES
The Nexus is not meant to be a universal replacement for all switches. Our belief is that you need to take a portfolio approach to find the correct switch for the particular needs of piece of your data center , so as cool as the Cisco Nexus 7000 is, you will still need rack switches, integrated blade switches, GE capability, 10GbE capability, etc. I actually believe this is a strong point for Cisco: there is a great amount of design flexibility across the Nexus 7000, Catalyst 6500, Catalyst 49xx, Catalyst Blade Switch 31xx/30xx, Cisco 7600, etc while still maintaining consistent management, operations, and features.

Also bear in mind that the Nexus is a new family–the Cisco Nexus 7000 is the first in the series, but there will be other form factors focused on other parts of the data center network.

Well, that’s about it–hopefully this adds to the discussion. Feel free to ping me if you have any questions.

Omar

]]>