Design for WAN Bonding, Speedfusion, PepVPN

Product Discussion Peplink Balance
1

Dear all friends and experts,

This is the 1st time, I work with Peplink products.

SpeedFusion_design_v01
SpeedFusion_design_v01826×627 42.8 KB

I have 03 branches ( 01 and 02 and 03 as network topology above )

I want to deploy WAN Bonding, (Speedfusion, PepVPN) for my company. here are all the devices I have for deploy.

Per each site/branch:

Peplink LB 2500 : 02 devices ( 8GE LAN, and 12 GW WAN)
Edge switch: 02 cisco 45 port GE
and core network infrastructure.

I want ask you guys some question about the design?

  1. How about HA between two LB 2500 ( LACP or another technology?
  2. I want to connect 08 LAN interfaces GE on LB-2500-01 to edge switch ( 4 interfaces to edge 01, and 04 to edge 02)
  3. I want to connect 08 LAN interfaces GE on LB-2500-02 to edge switch ( 4 interfaces to edge 01, and 04 to edge 02)
  4. I want to connect to 10 ISP per each LB 2500 ( LB-01 connect to 10 ISP, and LB-02 connect to 10 ISP).
  5. How about link aggregation between LB-2500 to edge switches?
  6. Please give me some advice for my physical design, after that I am going to config speedfusion to connect 03 site.

Thank you so much,
Lee Nguyen

2

Hi Lee
I see you have three ISPs Are all the ISPs links active all the time (or can they be)? Are they all sized the same or is the idea that the links on LB 2500 - 02 would be lower cost lower bandwidth links just used when there is an outage?

In answer to your questions:

  1. VRRP HA or consider a topology where you use both balance routers in active active, with half your remote sites connecting to one and the other half connecting to the other - with the remote sites failing over to the opposite core balance router if their primary one fails.
  2. OK
  3. OK
  4. No problem - what sort of links? What technology? What bandwidth?
  5. Of course using LACP.
  6. Although I seem to always be configuring it, apart from a very few situations where there is no other option, I don’t generally like the idea of devices in HA. It seems like an old fashioned way to approach high availability. Much prefer the remote branches to be aware of core device failure and then redirect traffic automatically.
3 Likes