VMware NSX :Underlying Gear & Physical Networking.

Hello Friends,

I am back on VMware NSX  track and This is my first post NSX. Now  going to do revised my NSX study, Hands-On and re-skill myself.   So I would love to hear you for any suggestion, correction that will help myself and other also.

Before begin want to give all credit to Blogger, VMware and Partner for their great write-up and sharing to enhance other knowledge also. 

It is my insight view on VMware NSX in reference of planning to deploy NSX for any Enterprise class environment.  Their few of requirement must to check and few of must to addressed, so here below those are:

  1. ESXi: Minimum Two (02) Host (ESXi) in each vSphere DRS cluster. Stop!!! Thinking because I am not saying that you cannot have a single standalone host within a cluster, absolutely its work but for HA (ESXi High availability) you have at least 02 Host in cluster.  I highly recommend to having three separate clusters with minimum three (03) Host.
  2. NSX License : NSX come with 3 different license editions; Standard, Advanced, and Enterprise. For more info check this NSX Licensing
  3. Max Minimum: It is bare minimum on the stand-alone Host clusters, the entire minimum to be one compute cluster, one edge and one management NSX. It is highly not recommended for any Production except PoC (Proof of concept).
  4. Network Minimum: With 1 Gig speed all just work well means if think that having 10 gigabit is a requirement for NSX that is not a true declaration.
  5. JUMBO Frame: Jumbo frames must be enabled and must to enabled on end to end communicated device like Physical switch and vNIC.
  6. TCP/IP Stack: TCP/IP stack got introduced in vSphere 5.5!!! Question what it does? Sure it is allow to create separate routing table for every different network that we create with default gateway per instance stack. So when you think of your default TCP/IP stack, you can think of things such as your Management kernel port, your vMotion kernel port, NFS, and iSCSI.
  7. VTEP & VXLAN: Cool VTEP is middle MAN who seat between L2 Fabric and Virtual Machine. VTEP remember connectivity flow of VMs with L2 logical networking. Its act as bridge.

Even though VXLAN is VLAN of VLANs to overcome of it limitation 4000 VLANs and its create logical VXLAN network through combing two or more VLANs. It support Layer 3 routing like OSPF, BGP & IS-IS, Also support multicast like IGMP & PIM.

Check this for more deep into VXLAN and VTEP

  1. VTEP HA : Now wish to share my environment which is running on a 1 gigabit fabric that have four NICs per host and I used two to have HA VTEPs, and two for HA management configuration. Let’s talk in details, if anyone running a 10G NIC configuration. I wish to have HA VTEPs however because it help to load balance my VXLAN tunnels between the two VTEPs on my hypervisor. So the recommendation is to have at least two…10 gigabit NIC.
  2. EDGE Racks: Yup!!! Most of flow going through edge racks even most of the tenants and customer connectivity will land here, so this major area there must have 10G speed and remain like management and Host networking OK with 1G speed.

Also shared L2 connectivity between management rack to edge rack to do fail over between the two and lets vSphere HA switch that.

  1. LEAF & SPINE: Not all but few folks are confuse between Leaf and spine, Let see what is Leaf and Spine. Leaf has access switches that connect to end point devices like servers, firewalls, load balancers, and edge routers etc. Spine nodes are aggregation nodes that provide connectivity between different racks in your datacenter.

Spine has leaf in it, like Network backbone where every Leaf switch is interconnected with each and every Spine switch. Leaf nodes are going to be your top-of-rack switch.

For more info click here Spin & Leaf

  1. STP: Do not use Spanning Tree Protocols (STP) not supported even not going to help because it is only help to avoid getting into any sort of networking loop.
  2. Layer 2 fabric: Let’s think about huge environment because Scale, so need to design carefully. L2 Fabric learn MAC address and use MAC to do communication and it can learn and save thousands or millions of MAC address entries.

Precisely you got it, also known as your ARP and MAC table.L2 Fabric has a limitation of just over 4000 VLANs in most of environments.

So I am going to take pause here and I’ll resume myself with next post soon. Once again Please do comment as your feedback and share your thought foe better understanding of mine and other also.

Thank you

Kuver P Singh

Advertisements

Lets meet at Delhi VMUG Event 23-Sep-17

Hello Friends,

Do register to your self for Delhi VMUG event, Announced date is Saturday, 23-Sep-2017.  It is first time in Delhi any event organized for local VMware user community.

So don’t missed it, You will have a chance to make new friends to expand your professional networking, you will learn about new technology such as NSX, vROPS & vRA and get free vSphere Tips & Tricks along with Career guidance.

Click here for Register yourself

If anyone failed to register or facing any issue. Please comment below this post.

Thanks to VMware VMUG Team @MyVMUG to design cool logo for Delhi_VMUG.

c678f0ba-bac4-4339-9879-5653d861ad02

Good Bye!!!Add-PSSnapin VMware.VimAutomation.Core

After up gradation of PowerCLi 5.x to 6.5, I got an error during execution of my old script “Add-PSSnapin : The Windows PowerShell snap-in ‘VMware.VimAutomation.Core’ is not installed on this computer.”

Finally got solution that I have to use “Get-Module –ListAvailable VM* | Import-Module” in place of “Add-PSSnapin VMware.VimAutomation.Core”.

After importing Module you can use all previous syntax.

Please see the below reference Release KB: PowerCli 6.5 Released Guide  and PowerCLi6.5 User Guide: PowerCli 6.5 User Guide

Feel free to comment to me for more conversation.

Thanks