BFD Session status Changes to Down & Up

While working on DataCom Network (composed of hundreds of NE routers, configured with MPLS, IS-IS, QoS, MPLS-VPN, MPLS-TE, BGP and list goes on and on……..!, I saw this alarm “BFD Session status Changes to down” and “BFD Session status changes to up”. After doing some research, studying and help, I came to know about the root cause and main reason for this alarm.

There can be different reason for alarm to appear:

1. The status of the interface goes down on which the BFD session is established between the two peers.

2. The BFD peer session is deleted or shutdown explicitly.

3. The link is not able to forward packets because of congestion or may be the link status is down.

In my case it was NSA (Non-Service Affecting) and I just need to confirm the status so for that we need to remember few commands which are:


display interface [interface name]

display bfd session all

There are bundle of more commands, if you need any help regarding that please feel free to reach me.

Note: If you are not familiar about BFD? Wait for my next blog on BFD

I Hope this will be informative for you :)

Multiprotocol Label Switching Fast Reroute (MPLS FRR)

In actual MPLS FRR is a feature of RSVP-TE, it is also called MPLS local restoration or MPLS local protection. MPLS FRR gives protection to the LSP path in a network where each LSP is protected by backup path. The node which redirects traffic after path failure to the backup path is known as Point of Local Failure (PLR) and the node where backup LSP merger with primary LSP is called Merge Point (MP). This protection is purely local as compare to protection which is enable at IP Layer (Layer-3) which take a bit more time then this which is not acceptable in real time application (VoIP, Video Conferencing etc). This local protection takes even less than 50 ms.

There are two types of protection approaches:

1. One-to-One Local Protection

In one-to-one approach, PLR maintain a separate path for each LSP across the path or network. This method creates a detour LSP for each protected path at the point of each local failure. For more detail, study RFC 4090

2. Many-to-One Local Protection

In this method, PLR create a single path that can be used to protect multiple LSP. In this method a same tunnel can be used for multiple LSP which acts as a protected path for all links which face failure. See RFC 4090 for more details.




Refer to the figure above, we have Node-A and Node-E as the start and end point for which the primary path (LSP) is from Node-A to Node-E through Node-B and Node-D. While the secondary path for Node-A to reach Node-E is through Node-C. We assume that for primary path FRR is enabled & once it is enable all the remaining nodes on a network came to know about this feature. Assume the link between Node-D and Node-E is down by any means, so the first node who realize this breakdown is Node-D which immediately inform Node-B and Node-A. For Node-A to get the failure message it will take some time and Node-D already know about the link failure & FRR is enable for the LSP so Node-D will use the detour path for the communication (Node-D-C-E) to avoid any loss and get rid of the link failure which will carry the traffic to final destination. This whole process will take less than 50 ms. On the other hand when the backup path (secondary LSP) comes up, traffic will be switched to the secondary LSP and detour path will turn down.

I Hope this will be informative for you :)

Ref: RFC4090

DHCP Relay

Dynamic Host Configuration Protocol (DHCP) serves to provide configuration parameters in a client/server environment where the DHCP server maintain address pool and lease information which the server used at time of requesting for IP from client side. This concept of DHCP relay is used when client and server does not reside on the same subnet. Look at the figure below; where the DHCP relay agent acts as a middle-man between host and DHCP server. When the host boots up on the network, the DHCP lease process occurs between the DHCP server and host (client).

DHCP lease process involves the following steps:

1. DHCPDISCOVER
In this step the host (client) sends a broadcast message over the network to request for IP address lease. The DHCP relay agent forwards the request to DHCP server after receiving from host.

2. DHCPOFFER
This can be a response from different server. This includes configuration parameters like IP, Lease information, MAC address, domain name etc as a unicast message to the host (client)

3. DHCPREQUEST
This is in response from the DHCP server to client who sends the initial DHCP server request, message means the client is requesting for IP address lease.

4. DHCPACK
This is sent by DHCP server to DHCP client which is the confirmation of assigning IP lease to the client.

As DHCPDISCOVER message is a broadcast over the network and you need a proper routing to route this message across the other segments where it has to be routed. For this you need to configure this on router interface so that it can forward your request to your DHCP server.

NOTE: In a routed network, you would need DHCP Relay Agents if you plan to implement only one DHCP server.



All clients on a network should be able to contact DHCP server. For this your server should be on the network topology and relied on by all TCP/IP based hosts with your environment. If your network is composed of different segments then we have to perform any of the following tasks:

a) Place a DHCP Relay agent on each segment
b) Place a DHCP server on each segment
c) Configure your router to forward a broadcast message


Configuration on Router

HUAWEI

You just need to configure a single command under the interface

[HUAWEI] interface GigabitEthernet 1/0/1
[HUAWEI-GigabitEthernet1/0/1] dhcp select {global | interface | relay}
[HUAWEI-GigabitEthernet1/0/1] dhcp select relay
[HUAWEI-GigabitEthernet1/0/1] ip relay address 192.168.20.20

Verification:

display dhcp relay statistics
display dhcp relay server {all | interface | vlan}

CISCO

Router # ip dhcp relay enable
Router # ip dhcp relay server 192.168.20.20 //Specify DHCP server address

Verification

Router# show ip dhcp relay conf
Router# show ip dhcp relay statistics

I Hope this will be informative for you :)

Network Design

Designing Network involves different steps, procedure and methods. I will talk about the first two section in this post, see other following post for the renaming:

1. Organizational Policies & Procedures
2. Essential of a Flexible Network
3. Network Design Method (PDIOO)
4. Design Process – 8 Steps

Organization Policy Cycle

An organizational policy and procedure is a collection of specific guidelines and rules in written form that are understood, implemented and maintained at every level of the organization for the purpose of reaching well-defined goals.



Guidelines for Organizational Model

Logical – Mirror the ecosystem.
Incremental – start small but think big
Horizontal – Get constant input from stakeholders
Manageable – Implement Control and access mechanism
Critical – Implement core application first
Exceptional – look for cutting edge solution


Network Infrastructure Essential

1. Availability: This means there should be Resiliency in your network, Redundant and 24/7 available
2. Efficiency: Best equipment, Services, Software, AAA, Queuing etc
3. Functionality: Like .net – so we have sufficient bandwidth for the application
4. Manageability: e.g. SNMP, Maintain performance, Security etc
5. Performance: you get what needed
6. Scalability: Future expansion

Designing: Access Layer – Distribution Layer & Campus Backbone

While preparing for my CCDA and CCDP, I came across with these information. I hope it will help you :)

Access Layer

1. Current & future needs for users or node ports
2. Can your company or client afford modular Cisco units
3. Is the existing cabling UTP adequate?
4. Can you afford to move to MM fiber?
5. Performance and Bandwidth requirement
6. Redundancy? & up to which level it is needed &/or Required
7. VLAN, VTP, STP or RSTP support requirement?
8. Layer-2 traffic pattern, Multicasting & QoS?

Distribution Layer

1. Layer-2 switch adequate? A Layer-3 switch?
2. Total user do you support? Or have to?
3. Do you need high availability?
4. Do you need distribution switches, modular & scalable?
5. What type of intelligence service like QoS, Security, IP Multicasting etc
6. Are you prepare for manageability and configurability
7. Are advance features need to be implemented like RSTP, MSTP, Backbone fast or Uplink fast.

Campus Backbone

1. Do you have three or more building connected through enterprise campus infrastructure?
2. Is your solution L2, L2/L3 or L3 throughout the network?
3. Are you read for high performance, multilayer switching?
4. Does your customer want to simplify and lower the number of links between distribution layer switches and server farm/edge distribution modules?
5. What are the performance needs?
6. How many high capacity links/ports do you need?
7. What are the high availability/redundancy needs?

I Hope this will be informative for you :)