IXPs reach 97 per cent of networks in Tajikistan but 21 per cent in Georgia

The share of networks that can exchange traffic through an internet exchange point (IXP) ranges from 97 per cent in Tajikistan to 21 per cent in Georgia across seven countries of the Caucasus, Central Asia and Türkiye, the Internet Society's Pulse tracker for October shows.

Pulse lists no active exchange point in Azerbaijan or Turkmenistan.

What the figures show

Pulse counts a network as covered if it is a member of an IXP or a customer of a member. The tracker uses PeeringDB, which it describes as based on self-reported data from IXPs and their members, and it lists only exchanges with three or more members. The measure shows reach, not traffic volume. The denominators also differ widely: one exchange can cover most of Tajikistan's 38 active networks, while Türkiye's figure is taken across 881.

• Tajikistan: 97 per cent of 38 active networks, through one exchange, TJ-IX in Dushanbe, with 21 members.
• Kyrgyzstan: 92 per cent of 61, through two exchanges, led by KG-IX in Bishkek with 22 members.
• Armenia: 75 per cent of 131, through ArmIX in Yerevan, with 34 members.
• Kazakhstan: 73 per cent of 218, through 14 exchanges.
• Türkiye: 63 per cent of 881, through six exchanges, including DE-CIX Istanbul with 70 members.
• Uzbekistan: 22 per cent of 108, through two exchanges.
• Georgia: 21 per cent of 152, through IXP.ge, with 14 members.

Gaps and discrepancies

Kazakhstan lists the most exchanges, but 12 of its 14 carry the KAZ-GOV-IX name, with nodes from Astana to Aqtau. Pulse puts Astana at 15 members and Almaty at 10, with the ten other nodes at between three and six. PeeringDB's facility record for KAZ-GOV-IX in Astana names JSC State Technical Service as the organisation. The other two exchanges, KazNIX and KAZ-IX, list 12 and six members. Pulse's headline gives 29 combined members for Kazakhstan, while its per-exchange counts sum to 83. The page does not explain the difference.

The same mismatch appears elsewhere. Türkiye's six exchanges list 121 members when their counts are added, against a stated total of 100, and Uzbekistan's stated total of 21 compares with 18 listed. A network that joins several exchanges could account for a lower total, but not for Uzbekistan's, where the stated figure is the higher one. Readers should treat the member totals as indicative and the coverage percentages as the tracker's headline measure.

Uzbekistan's 22 per cent rests on SNS-IX, which Pulse lists with 18 members, and UZ-IX, which it lists with none. PeeringDB's own record for UZ-IX shows five networks. Its Uzbek exchange listings are SNS-IX, UZ-IX and ITI-IX, and TAS-IX is not among them. The digital ministry says Uztelecom's new Bukhara and Kokand data centres connect to TAS-IX, the national internet traffic exchange point.

Georgia's 21 per cent rests on IXP.ge, which Pulse lists with 14 members at two physical locations. For Azerbaijan, Pulse's page says no active exchange is listed in PeeringDB. Absence from a voluntary database does not prove that no exchange operates, and Pulse excludes exchanges with fewer than three listed members.

Colocation listings

PeeringDB's public API, queried on 17 October, listed 49 facilities in Türkiye, nine in Armenia, seven each in Kazakhstan, Uzbekistan and Azerbaijan, five in Georgia, two in Kyrgyzstan, one in Tajikistan and none in Turkmenistan. All seven Azerbaijani listings are in Baku.

In Türkiye, the listing with the most networks present is MedNautilus Istanbul, listed under Telecom Italia Sparkle, with 50 networks, followed by Equinix IL2 in Istanbul with 48. The largest Kazakh listing, KazNIC in Semey, shows six.

The records count networks present, not megawatts or floor space, so they cannot be set against the capacity targets that governments publish. They also depend on operators registering themselves, which means a thin listing can reflect reporting habits as much as the market.