March 21, 2019 at 18:22Runet architecture
As our readers know, Qrator.Radar tirelessly explores the global connectivity of the BGP protocol, as well as the regional. Since the Internet is short for “interconnected networks” - “interconnected networks”, the best way to ensure high quality and speed of its operation is the rich and diverse connectivity of individual networks, whose development is primarily motivated by competition.
The fault tolerance of an Internet connection in any given region or country is related to the number of alternative routes between autonomous systems - AS. However, as we have repeatedly written in our studiesnational sustainability of global network segments, some paths become more important than others (for example, paths to Tier-1 transit providers or AS that host authoritative DNS servers) - this means that there are as many alternative routes as possible The bottom line is the only viable way to ensure system reliability (in the sense of AS).
This time, we take a closer look at the device of the Internet segment of the Russian Federation. There are reasons to keep our eyes on this segment: according to the data provided by the RIPE registrar database, 6183 AS out of 88,664 globally registered apply to the Russian Federation, which is 6.87%.
This percentage puts Russia in second place in the world in this indicator, immediately after the United States (30.08% of registered AS) and Brazil, which owns 6.34% of all autonomous systems. The effects that arise as a result of changes in Russian connectivity can be observed in other countries that are dependent on or adjacent to this connectivity and, finally, at the level of almost any Internet provider.
Diagram 1. Distribution of autonomous systems between countries in IPv4 and IPv6, top 20 countries
In IPv4, Internet providers from the Russian Federation announce 33933 of 774859 globally visible network prefixes, which represents 4.38% and puts the Russian Internet segment in fifth place rating. These prefixes, announced exclusively from the RU segment, cover 4.3 * 10 ^ 7 unique IP addresses out of 2.9 * 10 ^ 9 globally announced - 1.51%, 11th place.
Diagram 2. Distribution of network prefixes between countries in IPv4, top 20 countries
Under IPv6, Internet providers from the Russian Federation announce 1831 of 65532 globally visible prefixes, which represents 2.79% and 7th place. These prefixes cover 1.3 * 10 ^ 32 unique IPv6 addresses out of 1.5 * 10 ^ 34 globally announced - 0.84% and 18th place.
Diagram 3. Distribution of network prefixes between countries in IPv6, top 20 countries
One of the many ways to assess the connectivity and reliability of the Internet in a particular country is by ranking the autonomous systems belonging to a given region by the number of announced prefixes. This technique, however, is vulnerable to route disaggregation, which is gradually balanced by filtering excessively disaggregated prefixes on the equipment of Internet service providers, primarily due to the constant and inevitable growth of memory routing tables.
Table 1. AS size by the number of announced prefixes
We use the aggregated size of the announced address space as a more reliable metric for comparing the sizes of autonomous systems, which determines its potential and the limits to which it can be scaled. This metric is not always relevant in IPv6 due to existing IPv6 RIPE NCC allocation policies and protocol redundancy.
Gradually, this situation will be balanced by the growth in the use of IPv6 in the Russian segment of the Internet and the development of IPv6 protocol practices.
Table 2. AS size by the number of announced IP addresses
Both metrics - the number of announced prefixes and the aggregated size of address space - can be manipulated. Although we did not see similar behavior from the mentioned AS during the study.
There are 3 main types of relationships between autonomous systems:
Usually, these types of relationships are the same for any two Internet providers, which is confirmed in the region of the Russian Federation under consideration. However, sometimes it happens that two Internet providers have different types of relationships in different regions, for example, exchanging for free in Europe, but having commercial relations in Asia.
Table 3. AS connectivity by the number of customers The
number of customers of a given AS displays its role as a direct provider of Internet connectivity services to commercial consumers.
Table 4. AS connectivity by the number of peering partners
A large number of peering partners can significantly improve the connectivity of a whole region. Important here, although not required, are traffic exchange points (IX - Internet Exchange) - the largest Internet providers usually do not participate in regional exchanges (with a few noteworthy exceptions, such as NIXI) due to the nature of their business.
For a content provider, the number of peering partners can indirectly serve as an indicator of the volume of generated traffic - an incentive to exchange large volumes of it for free is a motivation factor (sufficient for most local Internet providers) to see a worthy candidate for peering partners in the content provider. There are also opposite cases when content providers do not support the policy of a significant number of regional connections, which makes this indicator not very accurate for estimating the size of content providers, that is, the volume of traffic generated by them.
Table 5. AS connectivity by the size of the client cone A
client cone is the set of all ASs that are directly or indirectly dependent on the autonomous system under consideration. From an economic point of view, each AS within a customer cone is, directly or indirectly, a paying customer. At a higher level, the number of AS within the client cone, as well as the number of direct consumers, is a key indicator of connectivity.
Finally, we have prepared another table for you that considers connectivity to the Runet kernel. Understanding the structure of the core of regional connectivity, based on the number of direct customers and the size of the client cone for each autonomous system in the region, we can calculate how far they are from the largest transit Internet providers in the region. The lower the number, the higher the connectivity. “1” means that for all visible paths there is a direct connection with the regional core.
Table 6. AS connectivity by distance to the core of regional connectivity.
What can be done to improve overall connectivity and, as a result, stability, reliability and security of any country, the Russian Federation in particular? Here are just a few of the measures:
The fault tolerance of an Internet connection in any given region or country is related to the number of alternative routes between autonomous systems - AS. However, as we have repeatedly written in our studiesnational sustainability of global network segments, some paths become more important than others (for example, paths to Tier-1 transit providers or AS that host authoritative DNS servers) - this means that there are as many alternative routes as possible The bottom line is the only viable way to ensure system reliability (in the sense of AS).
This time, we take a closer look at the device of the Internet segment of the Russian Federation. There are reasons to keep our eyes on this segment: according to the data provided by the RIPE registrar database, 6183 AS out of 88,664 globally registered apply to the Russian Federation, which is 6.87%.
This percentage puts Russia in second place in the world in this indicator, immediately after the United States (30.08% of registered AS) and Brazil, which owns 6.34% of all autonomous systems. The effects that arise as a result of changes in Russian connectivity can be observed in other countries that are dependent on or adjacent to this connectivity and, finally, at the level of almost any Internet provider.
Overview
Diagram 1. Distribution of autonomous systems between countries in IPv4 and IPv6, top 20 countries
In IPv4, Internet providers from the Russian Federation announce 33933 of 774859 globally visible network prefixes, which represents 4.38% and puts the Russian Internet segment in fifth place rating. These prefixes, announced exclusively from the RU segment, cover 4.3 * 10 ^ 7 unique IP addresses out of 2.9 * 10 ^ 9 globally announced - 1.51%, 11th place.
Diagram 2. Distribution of network prefixes between countries in IPv4, top 20 countries
Under IPv6, Internet providers from the Russian Federation announce 1831 of 65532 globally visible prefixes, which represents 2.79% and 7th place. These prefixes cover 1.3 * 10 ^ 32 unique IPv6 addresses out of 1.5 * 10 ^ 34 globally announced - 0.84% and 18th place.
Diagram 3. Distribution of network prefixes between countries in IPv6, top 20 countries
Individual size
One of the many ways to assess the connectivity and reliability of the Internet in a particular country is by ranking the autonomous systems belonging to a given region by the number of announced prefixes. This technique, however, is vulnerable to route disaggregation, which is gradually balanced by filtering excessively disaggregated prefixes on the equipment of Internet service providers, primarily due to the constant and inevitable growth of memory routing tables.
| | Top 20 IPv4 | | | Top 20 IPv6 | |
|---|---|---|---|---|---|
| ASN | AS Name | Number of Prefixes | ASN | AS Name | Number of Prefixes |
| 12389 | ROSTELECOM-AS | 2279 | 31133 | MF-MGSM-AS | 56 |
| 8402 | CORBINA-AS | 1283 | 59504 | vpsville-as | 51 |
| 24955 | UBN-AS | 1197 | 39811 | MTSNET-FAR-EAST-AS | thirty |
| 3216 | SOVAM-AS | 930 | 57378 | Rostov-as | 26 |
| 35807 | SkyNet-SPB-AS | 521 | 12389 | ROSTELECOM-AS | 20 |
| 44050 | PIN AS | 366 | 42385 | RIPN-RU | 20 |
| 197695 | AS-REGRU | 315 | 51604 | EKAT-AS | 19 |
| 12772 | ENFORTA-AS | 291 | 51819 | YAR-AS | 19 |
| 41704 | OGS-AS | 235 | 50543 | SARATOV-AS | 18 |
| 57129 | RU-SERVERSGET-KRSK | 225 | 52207 | TULA-AS | 18 |
| 31133 | MF-MGSM-AS | 216 | 206066 | TELEDOM-AS | 18 |
| 49505 | SELECTEL | 213 | 57026 | CHEB-AS | 18 |
| 12714 | Ti-as | 195 | 49037 | MGL-AS | 17 |
| 15774 | TTK-RTL | 193 | 41682 | ERTH-TMN-AS | 17 |
| 12418 | QUANTUM | 191 | 21191 | ASN-SEVERTTK | 16 |
| 50340 | SELECTEL-MSK | 188 | 41843 | ERTH-OMSK-AS | fifteen |
| 28840 | TATTELECOM-AS | 184 | 42682 | ERTH-NNOV-AS | fifteen |
| 50113 | SuperServersDatacenter | 181 | 50498 | LIPETSK-AS | fifteen |
| 31163 | MF-KAVKAZ-AS | 176 | 50542 | VORONEZH-AS | fifteen |
| 21127 | ZSTTKAS | 162 | 51645 | IRKUTSK-AS | fifteen |
We use the aggregated size of the announced address space as a more reliable metric for comparing the sizes of autonomous systems, which determines its potential and the limits to which it can be scaled. This metric is not always relevant in IPv6 due to existing IPv6 RIPE NCC allocation policies and protocol redundancy.
Gradually, this situation will be balanced by the growth in the use of IPv6 in the Russian segment of the Internet and the development of IPv6 protocol practices.
| | Top 20 IPv4 | | | Top 20 IPv6 | |
|---|---|---|---|---|---|
| ASN | AS Name | Number of IP Addresses | ASN | AS Name | Number of IP Addresses |
| 12389 | ROSTELECOM-AS | 8994816 | 59504 | vpsville-as | 2.76 * 10 ^ 30 |
| 8402 | CORBINA-AS | 2228864 | 49335 | NCONNECT-AS | 2.06 * 10 ^ 30 |
| 12714 | Ti-as | 1206272 | 8359 | Mts | 1.43 * 10 ^ 30 |
| 8359 | Mts | 1162752 | 50113 | SuperServersDatacenter | 1.35 * 10 ^ 30 |
| 3216 | SOVAM-AS | 872608 | 201211 | DRUGOYTEL-AS | 1.27 * 10 ^ 30 |
| 31200 | NTK | 566272 | 34241 | NCT-AS | 1.27 * 10 ^ 30 |
| 42610 | NCNET-AS | 523264 | 202984 | team host | 1.27 * 10 ^ 30 |
| 25513 | ASN-MGTS-USPD | 414464 | 12695 | DINET-AS | 9.51 * 10 ^ 29 |
| 39927 | Elight-as | 351744 | 206766 | INETTECH1-AS | 8.72 * 10 ^ 29 |
| 20485 | TRANSTELECOM | 350720 | 20485 | TRANSTELECOM | 7.92 * 10 ^ 29 |
| 8342 | RTCOMM-AS | 350464 | 12722 | RECONN | 7.92 * 10 ^ 29 |
| 28840 | TATTELECOM-AS | 336896 | 47764 | mailru-as | 7.92 * 10 ^ 29 |
| 8369 | INTERSVYAZ-AS | 326912 | 44050 | PIN AS | 7.13 * 10 ^ 29 |
| 28812 | JSCBIS-AS | 319488 | 45027 | INETTECH-AS | 7.13 * 10 ^ 29 |
| 12332 | PRIMORYE-AS | 303104 | 3267 | Runnet | 7.13 * 10 ^ 29 |
| 20632 | PETERSTAR-AS | 284416 | 34580 | UNITLINE_MSK_NET1 | 7.13 * 10 ^ 29 |
| 8615 | CNT-AS | 278528 | 25341 | LINIYA-AS | 7.13 * 10 ^ 29 |
| 35807 | SkyNet-SPB-AS | 275968 | 60252 | OST-LLC-AS | 7.13 * 10 ^ 29 |
| 3267 | Runnet | 272640 | 28884 | MR-SIB-MTSAS | 6.73 * 10 ^ 29 |
| 41733 | ZTELECOM-AS | 266240 | 42244 | ESERVER | 6.44 * 10 ^ 29 |
Both metrics - the number of announced prefixes and the aggregated size of address space - can be manipulated. Although we did not see similar behavior from the mentioned AS during the study.
Connectivity
There are 3 main types of relationships between autonomous systems:
- Client: pays another AS for traffic transit;
- Peer partner: AS exchanging its and client traffic for free;
- Supplier: receives traffic transit payments from other ASs.
Usually, these types of relationships are the same for any two Internet providers, which is confirmed in the region of the Russian Federation under consideration. However, sometimes it happens that two Internet providers have different types of relationships in different regions, for example, exchanging for free in Europe, but having commercial relations in Asia.
| | Top 20 IPv4 | | | Top 20 IPv6 | |
|---|---|---|---|---|---|
| ASN | AS Name | The number of customers in the region | ASN | AS Name | The number of customers in the region |
| 12389 | ROSTELECOM-AS | 818 | 20485 | TRANSTELECOM | 94 |
| 3216 | SOVAM-AS | 667 | 12389 | ROSTELECOM-AS | 82 |
| 20485 | TRANSTELECOM | 589 | 31133 | MF-MGSM-AS | 77 |
| 31133 | MF-MGSM-AS | 467 | 20764 | RASCOM-AS | 72 |
| 8359 | Mts | 313 | 3216 | SOVAM-AS | 70 |
| 20764 | RASCOM-AS | 223 | 9049 | ERTH-TRANSIT-AS | 58 |
| 9049 | ERTH-TRANSIT-AS | 220 | 8359 | Mts | 51 |
| 8732 | COMCOR-AS | 170 | 29076 | CITYTELECOM-AS | 40 |
| 2854 | ROSPRINT-AS | 152 | 31500 | GLOBALNET-AS | 32 |
| 29076 | CITYTELECOM-AS | 143 | 3267 | Runnet | 26 |
| 29226 | MASTERTEL-AS | 143 | 25478 | IHOME-AS | 22 |
| 28917 | Fiord-as | 96 | 28917 | Fiord-as | 21 |
| 25159 | Sonicduo-as | 94 | 199599 | Cirex | 17 |
| 3267 | Runnet | 93 | 29226 | MASTERTEL-AS | thirteen |
| 31500 | GLOBALNET-AS | 87 | 8732 | COMCOR-AS | 12 |
| 13094 | SFO-IX-AS | 80 | 35,000 | PROMETEY | 12 |
| 31261 | GARS-AS | 80 | 49063 | DTLN | eleven |
| 25478 | IHOME-AS | 78 | 42861 | FOTONTELECOM | 10 |
| 12695 | DINET-AS | 76 | 56534 | PIRIX-INET-AS | 9 |
| 8641 | NAUKANET-AS | 73 | 48858 | Milecom-as | 8 |
number of customers of a given AS displays its role as a direct provider of Internet connectivity services to commercial consumers.
| | Top 20 IPv4 | | | Top 20 IPv6 | |
|---|---|---|---|---|---|
| ASN | AS Name | Number of peer partners in the region | ASN | AS Name | Number of peer partners in the region |
| 13238 | Yandex | 638 | 13238 | Yandex | 266 |
| 43267 | First_Line-SP_for_b2b_customers | 579 | 9049 | ERTH-TRANSIT-AS | 201 |
| 9049 | ERTH-TRANSIT-AS | 498 | 60357 | MEGAGROUP-AS | 189 |
| 201588 | MOSCONNECT-AS | 497 | 41617 | SOLID-IFC | 177 |
| 44020 | CLN-AS | 474 | 41268 | LANTA-AS | 176 |
| 41268 | LANTA-AS | 432 | 3267 | Runnet | 86 |
| 15672 | TZTELECOM | 430 | 31133 | MF-MGSM-AS | 78 |
| 39442 | UNICO-AS | 424 | 60764 | TK-Telecom | 74 |
| 39087 | PAKT-AS | 422 | 12389 | ROSTELECOM-AS | 52 |
| 199805 | UGO-AS | 418 | 42861 | FOTONTELECOM | 32 |
| 200487 | FASTVPS | 417 | 8359 | Mts | 28 |
| 41691 | SUMTEL-AS-RIPE | 399 | 20764 | RASCOM-AS | 26 |
| 13094 | SFO-IX-AS | 388 | 20485 | TRANSTELECOM | 17 |
| 60357 | MEGAGROUP-AS | 368 | 28917 | Fiord-as | 16 |
| 41617 | SOLID-IFC | 347 | 31500 | GLOBALNET-AS | 14 |
| 51674 | Mehanika-as | 345 | 60388 | TRANSNEFT-TELECOM-AS | 14 |
| 49675 | SKBKONTUR-AS | 343 | 42385 | RIPN-RU | thirteen |
| 35539 | INFOLINK-T-AS | 310 | 3216 | SOVAM-AS | 12 |
| 42861 | FOTONTELECOM | 303 | 49063 | DTLN | 12 |
| 25227 | ASN-AVANTEL-MSK | 301 | 44843 | OBTEL-AS | eleven |
A large number of peering partners can significantly improve the connectivity of a whole region. Important here, although not required, are traffic exchange points (IX - Internet Exchange) - the largest Internet providers usually do not participate in regional exchanges (with a few noteworthy exceptions, such as NIXI) due to the nature of their business.
For a content provider, the number of peering partners can indirectly serve as an indicator of the volume of generated traffic - an incentive to exchange large volumes of it for free is a motivation factor (sufficient for most local Internet providers) to see a worthy candidate for peering partners in the content provider. There are also opposite cases when content providers do not support the policy of a significant number of regional connections, which makes this indicator not very accurate for estimating the size of content providers, that is, the volume of traffic generated by them.
| | Top 20 IPv4 | | | Top 20 IPv6 | |
|---|---|---|---|---|---|
| ASN | AS Name | Client Cone Size | ASN | AS Name | Client Cone Size |
| 3216 | SOVAM-AS | 3083 | 31133 | MF-MGSM-AS | 335 |
| 12389 | ROSTELECOM-AS | 2973 | 20485 | TRANSTELECOM | 219 |
| 20485 | TRANSTELECOM | 2587 | 12389 | ROSTELECOM-AS | 205 |
| 8732 | COMCOR-AS | 2463 | 8732 | COMCOR-AS | 183 |
| 31133 | MF-MGSM-AS | 2318 | 20764 | RASCOM-AS | 166 |
| 8359 | Mts | 2293 | 3216 | SOVAM-AS | 143 |
| 20764 | RASCOM-AS | 2251 | 8359 | Mts | 143 |
| 9049 | ERTH-TRANSIT-AS | 1407 | 3267 | Runnet | 88 |
| 29076 | CITYTELECOM-AS | 860 | 29076 | CITYTELECOM-AS | 84 |
| 28917 | Fiord-as | 683 | 28917 | Fiord-as | 70 |
| 3267 | Runnet | 664 | 9049 | ERTH-TRANSIT-AS | 65 |
| 25478 | IHOME-AS | 616 | 31500 | GLOBALNET-AS | 54 |
| 43727 | KVANT-TELECOM | 476 | 25478 | IHOME-AS | 33 |
| 31500 | GLOBALNET-AS | 459 | 199599 | Cirex | 24 |
| 57724 | DDOS-GUARD | 349 | 43727 | KVANT-TELECOM | 20 |
| 13094 | SFO-IX-AS | 294 | 39134 | UNITEDNET | 20 |
| 199599 | Cirex | 290 | 15835 | MAP | fifteen |
| 29226 | MASTERTEL-AS | 227 | 29226 | MASTERTEL-AS | 14 |
| 201706 | AS-SERVICEPIPE | 208 | 35,000 | PROMETEY | 14 |
| 8641 | NAUKANET-AS | 169 | 49063 | DTLN | thirteen |
client cone is the set of all ASs that are directly or indirectly dependent on the autonomous system under consideration. From an economic point of view, each AS within a customer cone is, directly or indirectly, a paying customer. At a higher level, the number of AS within the client cone, as well as the number of direct consumers, is a key indicator of connectivity.
Finally, we have prepared another table for you that considers connectivity to the Runet kernel. Understanding the structure of the core of regional connectivity, based on the number of direct customers and the size of the client cone for each autonomous system in the region, we can calculate how far they are from the largest transit Internet providers in the region. The lower the number, the higher the connectivity. “1” means that for all visible paths there is a direct connection with the regional core.
| | IPv4 top 20 | | | IPv6 top 20 | |
|---|---|---|---|---|---|
| ASN | AS Name | Connectivity rating | ASN | AS Name | Connectivity rating |
| 8997 | ASN-SPBNIT | 1.0 | 21109 | CONTACT-AS | 1.0 |
| 47764 | mailru-as | 1.0 | 31133 | MF-MGSM-AS | 1.0 |
| 42448 | ERA-AS | 1.0 | 20485 | TRANSTELECOM | 1.0 |
| 13094 | SFO-IX-AS | 1.0 | 47541 | VKONTAKTE-SPB-AS | 1.0 |
| 47541 | VKONTAKTE-SPB-AS | 1.07 | 13238 | Yandex | 1.05 |
| 13238 | Yandex | 1.1 | 8470 | MAcomnet | 1.17 |
| 3216 | SOVAM-AS | 1.11 | 12389 | ROSTELECOM-AS | 1.19 |
| 48061 | GPM-TECH-AS | 1.11 | 41722 | MIRAN-AS | 1.2 |
| 31133 | MF-MGSM-AS | 1.11 | 8359 | Mts | 1.22 |
| 8359 | Mts | 1.12 | 60879 | SYSTEMPROJECTS-AS | 1.25 |
| 41268 | LANTA-AS | 1.13 | 41268 | LANTA-AS | 1.25 |
| 9049 | ERTH-TRANSIT-AS | 1.16 | 44020 | CLN-AS | 1.25 |
| 20485 | TRANSTELECOM | 1.18 | 29226 | MASTERTEL-AS | 1.25 |
| 29076 | CITYTELECOM-AS | 1.18 | 44943 | RAMNET-AS | 1.25 |
| 12389 | ROSTELECOM-AS | 1.23 | 12714 | Ti-as | 1.25 |
| 57629 | IVI-RU | 1.25 | 47764 | mailru-as | 1.25 |
| 48297 | DOORHAN | 1.25 | 44267 | IESV | 1.25 |
| 42632 | MNOGOBYTE-AS | 1.25 | 203730 | SVIAZINVESTREGION | 1.25 |
| 44020 | CLN-AS | 1.25 | 3216 | SOVAM-AS | 1.25 |
| 12668 | MIRALOGIC-AS | 1.25 | 24739 | SEVEREN-TELECOM | 1.29 |
What can be done to improve overall connectivity and, as a result, stability, reliability and security of any country, the Russian Federation in particular? Here are just a few of the measures:
- Tax deductions and other benefits for local operators of traffic exchange points, as well as free access to them;
- Free or cheap land easement for laying fiber-optic communication lines;
- Conducting trainings and training sessions for technical personnel in remote regions, including workshops and other formats for teaching best practices for working with BGP. RIPE NCC organizes some of them, available by reference .
The data presented above is an excerpt from a study conducted by Qrator Labs regarding the second largest in the world regional Internet segment of the Russian Federation (also known as Runet) based on open data collected and processed as part of the Radar project . The presentation of the full study is announced as a workshop at the 10th Asia Pacific Regional Internet Governance Forum in July. A request for similar data for segments of other countries and regions can be sent to the e-mail address mail@qrator.net .