352-001 · Question #184
A data center provider has designed a network using these requirements: - Two data center sites are connected to the public Internet. - Both data centers are connected to different Internet providers.
The correct answer is C. One /20 block is used for the first data center and the second /20 block is used for the second D. One /20 block is used for the first data center and the second /20 block is used for the second. Splitting the /19 into two /20 blocks - one per data center - and advertising both the specific /20s and the aggregate /19 from both sites enables normal per-datacenter routing and automatic failover over the surviving ISP link.
Question
A data center provider has designed a network using these requirements:
- Two data center sites are connected to the public Internet.
- Both data centers are connected to different Internet providers.
- Both data centers are also directly connected with a private
connection for the internal traffic, and public Internet traffic can also be routed at this direct connection.
- The data center provider has only one /19 public IP address block.
Under normal conditions, Internet traffic should be routed directly to the data center where the services are located. When one Internet connection fails, the complete traffic for both data centers should be routed by using the remaining Internet connection. In which two ways can this routing be achieved? (Choose two.)
Options
- AThe data center provider must have an additional public IP address block for this routing.
- BOne /20 block is used for the first data center and the second /20 block is used for the second
- COne /20 block is used for the first data center and the second /20 block is used for the second
- DOne /20 block is used for the first data center and the second /20 block is used for the second
- EOne /20 block is used for the first data center and the second /20 block is used for the second
- FBGP will always load-balance the traffic to both data center sites.
How the community answered
(29 responses)- A17% (5)
- B3% (1)
- C69% (20)
- E3% (1)
- F7% (2)
Why each option
Splitting the /19 into two /20 blocks - one per data center - and advertising both the specific /20s and the aggregate /19 from both sites enables normal per-datacenter routing and automatic failover over the surviving ISP link.
An additional public IP address block is not required; the single /19 can be cleanly divided into two /20 sub-blocks to achieve per-site prefix advertising and failover without any additional address space.
This choice describes a prefix-splitting configuration that does not also include the aggregate /19 advertisement from both sites, so failover to the surviving ISP would not cover the full address range of both data centers.
Each data center advertises its own specific /20 prefix to its respective ISP, so under normal conditions BGP longest-prefix-match directs Internet traffic to the correct data center; both sites also advertise the /19 aggregate so the full address space remains reachable in both directions.
By having both data centers advertise the /19 aggregate to their ISPs, when one ISP connection fails the remaining ISP continues to carry the full /19 aggregate and routes all traffic for both data centers through the surviving connection, satisfying the failover requirement.
This choice represents a routing policy that does not simultaneously satisfy both the normal per-datacenter traffic steering requirement and the complete failover requirement when one ISP connection goes down.
BGP selects a single best path per prefix by default and does not automatically load-balance traffic; load balancing requires explicit multipath configuration and is not a built-in BGP behavior.
Concept tested: BGP prefix summarization and failover with /19 to /20 subnet splitting
Source: https://www.cisco.com/c/en/us/support/docs/ip/border-gateway-protocol-bgp/26634-bgp-toc.html
Topics
Community Discussion
No community discussion yet for this question.