Home
 Route Policy

The BGP Announcement Policy of Hong Kong Internet eXchange (HKIX) is as follows,

1. Filtering

a. IP address filtering

  • Applicable to general HKIX members.
  • Filtering by IP addresses.
  • Update via Filter Update Agent for each announcement.

b. AS number filtering

  • Applicable to HKIX members who have more than 5 Autonomous Systems (AS) connected and each AS contains at least one /20 or shorter prefix announced to the HKIX.
  • Filtering by AS number orignated from the IAP's customer network.
  • Update via Filter Update Agent for each announcement.
  • Separate application is required, please make your application by e-mail to hkix-noc@cuhk.edu.hk.

HKIX members should check the integrity of the filters and submit only the IP address or AS number that require for update. HKIX may reject the applications if the IP addresses or AS numbers had already included in the filter.

The minimum IP block accepted by the HKIX route server is /24.

2. Non-filtering

Closer Peering Arrangement (CPA) Scheme

  • Applicable to HKIX members who are using AS filtering and advertising more than 2500 prefixes for 200+ different ASes to the HKIX.
  • All routes will be accepted by the route servers.
  • No need to submit filter update.
  • Separate application is required, please read here for details.

3. IPv6 BGP Filtering

HKIX have implemented bogon filter for IPv6 address filtering. In order to facilitate better routing management for IPv6 routes being advertised via HKIX/AS4635 MLPA Route Server, we highly recommend all participants to make use of BGP community tagging when they announce/receive IPv6 routes to/from the HKIX MLPA route server.

The BGP communities supported by HKIX MLPA route server (for IPv6 only)

1
4635:900

Upstream / peer routes announced by free IPv6 transit providers for use by those HKIX participants which seek for free IPv6 transit over HKIX MLPA

2
4635:800
Downstream / internal routes seeking for peering only over HKIX MLPA but not for transit
3
4635:700
Downstream / internal routes seeking for transit over HKIX MLPA

HKIX participants should tag the routes they announce to HKIX MLPA route server as follows:

1. For HKIX participants which want to provide free IPv6 transit to other HKIX participants over HKIX MLPA:

  • They should announce their upstream / peer routes with tag 4635:900
  • They should announce their downstream / internal routes with tag 4635:800.

2. For HKIX participants which want to do multilateral peering over HKIX only:

  • They should announce their downstream / internal routes with tag 4635:800
  • They should NOT announce their upstream / peer routes.

3. For HKIX participants which seek for transit over HKIX MLPA:

  • They should announce their downstream / internal routes with tag 4635:700
  • They should NOT announce their upstream / peer routes.

 

HKIX participants should process the routes learned from HKIX MLPA route server as follows:

1. For HKIX participants which want to provide transit to other HKIX participants over HKIX MLPA:

  • When they see routes with tag 4635:900, they should just ignore them
  • When they see routes with tag 4635:800, they should re-distribute the routes to their downstream / internal networks only, but not to their upstream / peer networks
  • When they see routes with tag 4635:700, they should re-distribute the routes to all their downstream / internal / upstream / peer networks.

2. For HKIX participants which want to do multilateral peering over HKIX only:

  • When they see routes with tag 4635:900, they should just ignore them
  • When they see routes with tag 4635:800 and 4635:700, they should re-distribute the routes to their downstream / internal networks only, but not to their upstream / peer networks.

3. For HKIX participants which seek for transit over HKIX MLPA:

  • They should re-distribute all the routes that they learn from HKIX MLPA route server to their downstream / internal networks only, but not to their upstream / peer networks.

 

 

 

Copyright 2006. The Chinese University of Hong Kong. All rights reserved.

Site map Home Contact Us