With Marcos Hernandez
Welcome to the Cisco Support Community Ask the Expert conversation. This is an opportunity to learn about FlexPod which is a predesigned, base configuration that is built on the Cisco Unified Computing System (Cisco UCS), Cisco Nexus data center switches, NetApp FAS storage components, and a range of software partners. Marcos Hernandez is a technical solutions architect in the Data Center Solutions Go-to-Market Team, with over 16 years of professional experience and 11 years at Cisco. He currently leads the technical partner enablement efforts for FlexPod, covering Cisco Unified Computing System, Cisco Nexus, and NetApp products. He is also responsible for the technical enablement for cloud and automation solutions for the channel. His responsibilities include training and technical mentoring as well as field awareness and communication.
Remember to use the rating system to let Marcos know if you have received an adequate response.
Marcos might not be able to answer each question due to the volume expected during this event. Remember that you can continue the conversation on the Data Center sub-community discussion forum shortly after the event. This event lasts through December 9th, 2011. Visit this forum often to view responses to your questions and the questions of other community members.
Can you summarize the capabilities of the remotely accessible demo and training resources that Cisco has made available to partners? Also, who has access to them? Are there geographical limitations? Is there a cost?
Thank you for your questions Mark.
Cisco and NetApp are very excited to announce the availability of a new, first-of-its-kind, training and demonstration capability around our joint datacenter solution, FlexPod. As mentioned earlier, there are two offerings being announced:
1) FlexPod Implementation Best Practices Lab: The FlexPod Lab is a new resource that allows Cisco and NetApp Partners to become intimately familiar with a FlexPod implementation in the context of a Virtual Desktop (VD) deployment.
2) FlexPod Demonstration Resources: The FlexPod Demonstration Lab is a new resource that allows Cisco and NetApp Partners to conduct demonstrations that seek to highlight the key value proposition of FlexPod and its business transformation advantages, in the context of a Virtual Desktop Infrastructure (VDI) deployment.
Both resources are available to all certified Cisco/NetApp partners. The program is open to these partners on a worldwide basis and it is FREE OF CHARGE.
We have paid great attention to usability and ease-of-use. To access the training labs all you need is a laptop running an Internet browser (IE, Safari, Chrome, etc.). The entire experience takes place inside this browser session, using Virtual Desktop access against a hosted management station. A simple and elegant process.
At this time, twelve (12) training PODS and two (2) demo PODS are being made available, with more to be added in the future depending on the success and penetration of the program. The access is offered in two ways: 1) a first-come, first-serve basis and 2) a proctored instructor led training (for the FlexPod Implementation lab only).
To access the Demo and/or the Implementation Labs, follow these steps described on:
The Demo/Lab Program overview can be found on:
I hope this answers your questions. We encourage all of our joint partners to sign up for the labs and start building their technical skills ASAP.
is it possible to see somewhere on UCS what are the current IOPS numbers for the storage?
Measuring IOPS is typically a function found on the OS/Hypervisor side via an agent, or the storage side. UCS Manager offers some average packet counts in the stats section (this is not related to IOPS) and also the ability to enable traffic monitoring sessions via SPAN for Fibre Channel sources, specifically:
• Uplink Fibre Channel port
• SAN port channel
• Service profile vHBA
• Fibre Channel storage port
Notice that you will still need a FC traffic analyzer. For information on how to configure this monitoring session, take a look at Chapter 39 of the 2.0 UCS CLI Configuration Guide, or Chapter 41 of the GUI Configuration Guide on:
Here is a screenshot of the FI stats available through UCSM:
In the specific case of FlexPod, we offer a performance management tool by NetApp called the NetApp Management Console (NMC). This tool allows you to monitor, in real time, the storage IO performance of the entire stack, of a specific volume or for an individual vFiler. It is really powerful and we provide a very comprehensive demonstration of its capabilities in Demo task 3.2 of the business relevance demo script: Uncompromised User Experience - NetApp Flash Cache. The demonstration script document can be found here. Here is a screenshot of the type of information you will see:
You can try using the demonstration PODS to see it for yourself by visiting:
The Program overview can be found on:
Thanks for the answer Marcos!
I must admit I'm not a storage guy and still in process of learning the terminology. I can see some Rx packets and Tx packets in UCSM or latency in NMC but I'm not sure how do you calculate IOPS from it. For example, if I know that some VM will require around 50 IOPS, I would like to know what my system is capable of and how many VMs I'm allowed to create.
I should be more specific here. While important, the "transport" of SCSI traffic (e.g. Fibre Channel) does not directly affect the calculation of IOPS, which is a function of the storage system, hence you have to measure it there, or on the host side via agent (performance tab in vCenter, if you are using vSphere OS, for example).
The screenshot I posted for the NMC is just one example of the many things that can be tracked by NetApp's performance advisor. You need, of course, to be using a NetApp storage product to leverage this tool.
I hope this helps.
On shared fabrics such as #flexpod, QOS is paramount. Other than egress QOS on UCS why is there no references for storage QOS 'end-to-end' on #flexpod. NetApp native COS 4 markings on ethernet and QOS mutation difficulties on N7K-Q1/F1 blades / N5K, means no reference model we can get our teeth in to! We had to build our own (including weightings / pause frames / no-drop) on the network fabric (to match the egress UCS QOS) whereupon LLDP to pure IP storage (NFS) on NetApp CNA's is not even supported unfortunately
This is a shame for customers in congestion scenarios / high IOPS environments as only supporting PFC for FCOE (whereupon DCBx/COS markings are pure Layer2/Ethernet constructs anyway?) means we cannot emulate FC no-drop storage on a pure IP/NFS storage environment (which is where most VSphere clusters are going due to benefits of NFS).... any thoughts on why the CVDs don't address QOS fully?
Donal, you are right in that the CVD currently only addresses FCoE no-drop class (hardcoded QoS group) in the Nexus fabric. I have seen countless internal threads discussing the pros and cons of no-drop queuing for NFS traffic. I am of the opinion that NFS (and iSCSI for that matter) will NOT benefit from PFC, due to the very nature of TCP. I would not put TCP traffic in a lossless class. Additionally, what is well-known for FCoE in terms of PFC (marking and control between host port and switch access port) is not so much for other types of storage traffic, meaning you would have to do all this manually (I guess you are saying that you did this already).
The BU folks recently updated the QoS best practices for all products running the NX-OS stack, and the conclusion is that NFS will be better served by a traditional TCP queuing mechanism, such as Priority Queuing/Class-Based Weighted Fair Queuing/Weighted Round Robin (PQ/CBWFQ/WRR), which can be applied starting at the N1KV (our dVS). Some fine tuning might be necessary. IMPORTANT: Not all Nexus products support all queuing implementations. Check the feature reference guide for details.
Here is an abstract from their design doc. It is a little cryptic, but you will see where they are going with this:
Data Optimization (NFS/iSCSI)
What is the goal ? What is the traffic?
Final comment on QoS: in the presence of real time traffic (VoIP, Video), I would not apply PQ to NFS.
I will bring this to the attention of the CVD owners and urge them to cover QoS in more detail in upcoming versions of the document.
I hope this helps,
I have a quick question for you. Can multiple FlexPod training sessions occur at the same time? If not, how is potential conflict handled?
The answer is YES, multiple trainings can occur at the same time. At this moment, 12 training PODS are being made available with more to be added in the future depending on the success and penetration of the program. The access is offered in two ways: 1) a first-come, first-serve basis and 2) a proctored instructor led training, which can only be approved by Cisco and NetApp. When no sessions are available, participants are waitlisted and informed about last minute cancellations.
Thanks a lot,
Here are some questions I hacve in regards the FlexPod:
Thank you for your responses. John
Thanks for your questions.
Details on how to prepare for the training, where to obtain the accompanying collateral and useful multimedia content can be found on the FlexPod Community page at:
For issues while preparing for a training, or while delivering a training, please email the program administrators between 7 am PST and 6 pm PST:
You can also post your inquiry on the FlexPod Community page, found at:
The training steps are very prescriptive and predictable and the infrastructure is extremely stable, but if a problem were to arise, users can revert the environment back to a good-known starting point by clicking on the “reset” button located on the left control panel.
I want to better understand the FlexPod. Here are few questions:
Thank you for your responses.
At this point, all collateral and tools are offered only in English.
For comments, questions and feedback please email the program administrators at:
You can also post your inquiry on the FlexPod Community page, found at:
The FlexPod training lab program is available NOW. Please come check it! :-)