Designing BGP routing with BGP attributes to influence the traffic flows based on the desired traffic patterns (load sharing, active/passive)

What are the main BGP attributes you can manipulate to influence BGP path selection?

The main BGP attributes that can be manipulated include Local Preference, AS Path Prepending, Multi-Exit Discriminator (MED), and Community attributes. Local Preference determines outbound traffic preferences, AS Path Prepending artificially lengthens the AS Path to make a route less preferable, MED influences inbound traffic from the same neighboring AS, and Community attributes can be used to apply routing decisions across multiple routers.

How does changing the BGP Local Preference attribute affect traffic flow?

Changing the BGP Local Preference affects outbound traffic flow because it indicates to the BGP process which path is more preferred when multiple routes to the same destination exist. A higher Local Preference value is more preferred than a lower one. By setting a higher Local Preference on one path, traffic is influenced to exit through that path.

How can AS Path Prepending be used for load sharing?

AS Path Prepending can be used for load sharing by extending the AS Path length of one or more routes to influence incoming traffic to prefer shorter AS Paths. By prepending extra AS numbers to the path, that particular path becomes less desirable, causing traffic to be distributed over other available paths with shorter AS Paths, thus achieving a form of load sharing.

What is the role of the Multi-Exit Discriminator (MED) in controlling BGP traffic patterns?

The Multi-Exit Discriminator (MED) is used to signal to external neighbors the preferred path into an AS when multiple entry points exist. It is used for inbound traffic control. Lower MED values are more preferred, so by setting a lower MED on one entry point, traffic from the neighboring AS is more likely to take that path.

Can you explain the difference between Active/Passive and Load Sharing traffic patterns in BGP design?

Active/Passive traffic patterns involve having one primary path for traffic and a backup path that is only used if the primary fails, which is typically used for redundancy. Load Sharing, on the other hand, seeks to distribute traffic across multiple paths simultaneously to utilize available bandwidth more efficiently and avoid overloading a single link.

In what situation would you prefer to use BGP communities over other attributes?

You may prefer to use BGP communities when you need to apply consistent routing policies across multiple routers without changing the fundamental path attributes like AS Path or Local Preference. BGP communities can tag routes and signal to community-aware routers to take specific actions like accept, prefer, or suppress routes across a wide network region.

How can you implement an Active/Passive setup with BGP attributes to ensure redundant connectivity to your network?

To implement an Active/Passive setup, you can utilize a combination of attributes like AS Path Prepending, Local Preference, and MED. By setting a higher Local Preference and lower MED for the active path, and prepending additional AS numbers to the passive path, you ensure that the active path is always preferred under normal circumstances and the passive path is only used if the active path becomes unavailable.

What impact does setting a high Local Preference on all routers in an AS have when there are multiple exit points?

Setting a high Local Preference on all routers within the same AS will not help in determining which exit point routers should use since Local Preference is only significant within an AS. For influencing decision-making between multiple exit points, one would need to use attributes like MED or weight, which are evaluated at the border routers.

Why might you want to lower the Local Preference for a route, and what would be the result?

Lowering the Local Preference for a route may be desirable to make that route less preferred and thus influence the BGP process to choose a different path for outbound traffic. The result would be traffic avoiding the path with the lowered Local Preference when other more preferred paths are available.

What are the potential risks of using BGP AS Path Prepending excessively, and how might you mitigate them?

Excessive AS Path Prepending can lead to routing instability and sub-optimal paths. Long AS Paths can cause routers to drop updates due to policy restrictions or path size limitations. To mitigate these risks, ensure that prepending is done with a clear understanding of the routing policies of all intermediary ASes and limit the number of prepends to the minimum necessary to achieve the desired traffic pattern.

In AWS, how can you manipulate BGP attributes to prefer one Direct Connect connection over another for incoming or outgoing traffic?

In AWS, you can manipulate BGP attributes on Direct Connect connections by setting a lower MED to make one connection more preferable for incoming traffic from AWS. For outgoing traffic, you can set a higher Local Preference on the preferred path within your network to influence which Direct Connect connection is used.

How do you configure BGP to failover to another path when the primary path becomes congested, not just when it goes down?

BGP itself does not inherently detect or react to congestion; it makes decisions based on path attributes. To implement failover on congestion, you would need an additional mechanism, like Performance Routing (PfR) or Software-Defined WAN (SD-WAN) technologies, which can monitor live traffic conditions and adjust routing. These technologies can work with BGP, adjusting its attributes like Local Preference or MED when predefined thresholds are reached, to shift traffic away from congested paths.