Dell PowerEdge MX7000 server Owner's Manual

Add to my manuals
3 Pages

advertisement

Dell PowerEdge MX7000 server Owner's Manual | Manualzz

PowerEdge Product Group

Direct from

Development

PowerEdge MX7000 Chassis

Management Networking Cabling

Tech Note by:

Brian Claggett

Michael E. Brown

New Feature: Automatic Uplink Detection

The first thing to notice about the MX7000 network physical chassis is labelling: as opposed to the older M1000e chassis, you will notice that the management ports are not label led “STK” (stacking) or GB (uplink), but rather just GB 1 and

GB 2. Because of the new automatic detection features in MX7000, you can use any port for stacking and/or uplink, and the management module will automatically detect each port to determine if it is a stacking or uplink port.

Figure 1 – MX7000 Management Module

SUMMARY

The purpose of this whitepaper is to describe the new and improved features of the

MX7000 chassis management network, provide recommended network cabling diagrams, and show the fault tolerant capabilities of the new network design in the new Dell EMC 14G

MX7000 chassis.

The MX7000 chassis features dual redundant management modules, with each management module featuring two management network ports, for a total of 4 management network ports on the chassis.

The management network is meant to provide network connections for chassis management that is separate from the customer data network.

There are several new and improved design points for this network compared to previous generation chassis that significantly impact how these new chassis should be cabled and managed.

New Feature: Network Loop Prevention

When cabling chassis in a stack, you will notice is what appear to be connections forming network loops and redundant connections between the chassis. The management modules will automatically detect redundant ports between chassis and will block packets automatically to prevent network loops from forming.

There are a couple points worth noting about this feature. First, when you stack chassis and cable them together, only one uplink will ever forward traffic out of the stack. The recommended cabling configuration has two links from chassis to the top of rack switch. However, if you have the recommended redundant connections out of the stack, all of the “extra” redundant connections will be in a

“blocking” state, i.e. Link-UP, but not forwarding network traffic. If the system detects a link failure of the active link, the chassis will automatically fail over to the redundant links.

Cabling an Individual Chassis

When cabling an individual chassis, connect one network cable from each management module to the data center top of rack switch. Ensure that both ports on the top of rack switch are enabled and on the same network and

VLAN.

Figure 2 – Schematic representation of single chassis connection

© 2018 Dell Inc. or its subsidiaries. All Rights Reserved. Dell, EMC and other trademarks are trademarks of Dell Inc. or its subsidiaries

.

This configuration provides redundant network access to the chassis, protecting against link failure or management module failure. You can cable the redundant connections to the same top of rack switch or to two redundant switches, just be sure to configure both connections the same and have them on the same network and VLAN.

As mentioned above, there is no need to prefer either GB 1 or GB 2 for these connections.

Cabling an Individual Chassis

With the automatic uplink detection and network loop prevention, it is possible to connect multiple chassis one to the other with cables to save port usage in the data center switches and still access each chassis over the network. When multiple chassis are cabled in this manner it is known as a “stack”.

Note: While the automatic uplink detect and network loop prevention allow for chassis to be interconnected without network issues, stacked chassis are not automatically part of a management group .

See the MX7000 Multi Chassis Management feature for group management. This feature (not covered in this whitepaper) lets you manage multiple chassis through one management module, and requires the network stacking arrangement described here.

Recommended Topology for Multiple Chassis

While the auto sensing loop prevention algorithms allow for virtually any topology and provides access to all components in the stack of chassis, there are optimum topologies for connecting a group of chassis. These topologies provide redundant connections into the stack and optimize network down time for firmware updates and possible faults. The cabling diagram shown will protect against any single point of failure in the entire stack. With this configuration you can survive the following failures and retain full management network access to all nodes in the stack:

Figure 3

– Schematic representation of multiple chassis connections

Any single network cable failure

Any single management module failure

Power loss to any single chassis in the stack

It is important to note that these are the minimum failures that we have designed protection for. It is possible to have more failures than the above and still have full functionality, but it would depend heavily on where those failures occurred.

Here is an example topology for a 12

Chassis Dual MM chassis stack, connected to a top of rack switch at ports 1 and 4:

For consistent access to all chassis and their components from a management network, it is important that both ports 1 and 4 are connected to the same layer 3 network. A very common misconfiguration that we have seen is accidentally connecting the two redundant uplink ports to different VLANs. If the chassis are

© 2018 Dell Inc. or its subsidiaries. All Rights Reserved. Dell, EMC and other trademarks are trademarks of Dell Inc. or its subsidiaries

PowerEdge Product Group

Direct from

Development

connected to different VLANS, when the primary link fails, you will lose access to your management network.

As can be seen, all chassis are connected with their neighbor chassis with redundant connections. The two chassis that are connected to the top of rack switch are connected to each other, providing a redundant connected path. This redundant path is important to allow the stack to survive power loss to a full chassis in the middle of the stack.

While the management module contains a concept of “Active” versus “Passive” roles, networking for the chassis’ is independent of these roles.

The networking algorithm uses Spanning Tree Protocol to determine a network path to each and every component in the

FabD stack. This may include forwarding paths through “Passive” MM modules. It is important to note that Spanning

Tree is *not* enabled on any uplink ports to top of rack switches. Spanning Tree is automatically configured between

MX7000 chassis, and is not ever configured on links that are not MX7000 chassis.

One important thing to be aware of are network disruptions caused by cabling or link changes in the stack. When links change, for example if a cable is pulled or added, the network may go down for approximately 30 seconds to re-compute the network topology. This affects all chassis in the stack. One thing you will notice is that every time cabling or link state changes in the stack, all iDRACs and management modules in the stack will reconfigure their network. If the components are configured for DHCP, you will see them refresh their IP addresses.

The above is true for both DHCPv4 and DHCPv6.

Reserved VLANS

To implement the features for auto uplink detection and network loop prevention, the MX7000 chassis internally use

VLAN in the series 4000 through 4020. This series is reserved by the management module and chassis components and cannot be used for access through the uplink connection.

All other possible VLANS are available for use and VLAN tagged packets will forward through the stacked chassis.

MX7000 External Port Settings

For all chassis in a stack, it is important that all chassis external port settings are set to the same configuration.

Each management module in the MX7000 chassis has two RJ45 Ethernet ports labeled “gb1” and “gb2”. Each port is configured to be Auto MDI-X, which typically eliminates the need for an Ethernet crossover cable when connecting any

MX7000 port to another chassis or top of rack switch. This is a non-configurable setting.

Some port settings, such as speed and auto-negotiation can be configured through the management GUI. All ports on the chassis will configured with the selected values. When actively changing this chassis setting, there will be a delay in access while the settings are applied.

Note: If non-auto negotiate settings are selected, the switch that the chassis is connected with MUST be also changed to the same speed and duplex values.

Additionally, Auto MDI-X may not function properly when autonegotiate is off, resulting in a link down state.

In a dual management module chassis, all four external gb1 and gb2 ports are set to the same settings.

Figure 4

– Management Module Web Page Port Settings

© 2018 Dell Inc. or its subsidiaries. All Rights Reserved. Dell, EMC and other trademarks are trademarks of Dell Inc. or its subsidiaries

advertisement

Was this manual useful for you? Yes No
Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Related manuals

Download PDF

advertisement