NSX Useful numbers – VCP-NV Study

These are some of the useful numbers I’ve collated as I’ve been doing the study guide. I’m happy to add more if anyone thinks I’ve missed some important ones.

Prerequisites

VMware vCenter Server 5.5 or later

VMware ESX 5.0 or later for each server

VMware Tools – For vShield Endpoint and NSX Data Security, you must upgrade your virtual machines to hardware version 7 or 8 and install VMware Tools 8.6.0 released with ESXi 5.0 Patch 3.

MTU minimum 1550, 1600 recommended

vHardware Minimums

NSX Manager NSX Controller Edge Compact Edge Large Edge Quad Large Edge X-Large vShield Endpoint NSX Data Security
vCPU 4 4 1 2 4 6 2 1
vMEM 12Gb 4Gb 512Mb 1Gb 1Gb 8Gb 1Gb 512Mb
vDISK 60Gb 20Gb 512Mb 512Mb 512Mb 4.5Gb
(4Gb swap)
4Gb 6Gb

Maximums

DHCP Pools – 20,000

Segment IDs (VNI) – 10,000

Audit Logs – 1,000,000

TCP Ports

443 between the ESXi hosts, vCenter Server, and NSX Manager.

443 between the REST client and NSX Manager.

TCP 902 and 903 between the vSphere Web Client and ESXi hosts.

TCP 80 and 443 to access the NSX Manager management user interface and initialize the vSphere and NSX Manager connection.

TCP 22 for CLI troubleshooting.

vSphere Scalability

1000 Hosts

10000 VMs

10000 Port groups

60,000 virtual ports

Logical Router

up to 1000 interfaces

up to 8 uplinks

1200 DLR instances total

100 DLR instances per ESXi host

Edge

HA default heartbeat failover interval 15 seconds (5x 3 seconds), can be reduced to 6 seconds minimum.
HA heartbeat minimum frequency = 1 second

NSX Manager

1 NSX to 1 vCenter

Backup Frequency – Hourly/Daily/Weekly

NSX Controllers

3 minimum, must be an odd number

VXLAN

VNI – 24bit number – 16M VXLAN networks

MTU 1600 bytes to support VXLAN with IPv4 and IPv6 traffic

Allows for:
50 bytes overhead for VXLAN encapsulation
54 bytes overhead if VLAN tagging is used

Uses UDP port 8472 as a transport protocol

LACP

64 LAGs per host

64 LAGs per distributed switch

802.1Q – VLAN tagging

4096 VLANs maximum

VPN

IPsec VPN – maximum of 64 tunnels across maximum of 10 sites

SSL VPN – up to 25 users

9 thoughts on “NSX Useful numbers – VCP-NV Study

  1. I cannot find any documentation to reference the Edge HA heartbeats. Do you have the reference? All other sources seems to indicated for a 15 s deadtime, the frequency is 3 secs instead of 5 secs.

    HA default heartbeat failover interval 15 seconds (3x 5 seconds), can be reduced to 6 seconds minimum (3x 2 seconds)
    HA heartbeat frequency = 1 second

    • It’s a long time ago, but I suspect I was going on what I was told on a course.

      Default output from the Edge CLI command “show service highavailability” shows:
      Unit Poll Policy:
      Frequency: 3 seconds
      Deadtime: 15 seconds
      I don’t have access to an NSX environment at the moment to test the effects on the frequency of changing the dead time unfortunately.

  2. Yes, the default dead time is 15sec (can be set as low as 6 seconds) and the freq 3 sec.
    https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2126560

    Also, just fyi b/c I dont think it will be on the test but changes in 6.2.3 for reference
    “New VXLAN Port 4789 in NSX 6.2.3 and later: Before version 6.2.3, the default VXLAN UDP port number was 8472. See the NSX Upgrade Guide for details” Also some changes in DHCP sizes for Edge
    http://pubs.vmware.com/Release_Notes/en/nsx/6.2.3/releasenotes_nsx_vsphere_623.html

Leave a comment