The impact of “emergencies” on the relocation and availability of web resources

The Impact of War in Ukraine on Relocation and Availability of Web Resources

I share my personal observations about the current situation with Internet resources in Ukraine. It is possible that this information will be interesting or useful for those who are engaged in IT in order to see how the situation is roughly developing in numbers, to understand behavioral models, volumes, directions of migration, and to assess risks.

On the infographic, I even depicted the situation in excessive detail, but below I will describe in more detail what the data came from, the research method, errors, and some conclusions.

Indicators

Source Collection

of web resources is represented by 1571 domains that were hosted on the territory of Ukraine at the time of January 2022 and another 74634 domains from 10 countries I selected for comparative analysis (of course, this is not an exhaustive list for each country, but only the one that was at my disposal before the start of all events ; in general, it also reflects the approximate share of each in the total contribution to the “providing” of resources, since the collection was carried out on equal terms).

The original collection was interviewed and divided into 4 groups:

  1. moved outside – the number of resources that have changed the hosting country (the “moved outside” summary contains brief moving statistics);
  2. Unavailable – the number of resources for which the poll failed;
  3. moved inside – number of resources that have changed ip address and location/city within the country;
  4. stay – the number of resources that have not significantly changed their location and availability status (the ip address could change, but the city of location could remain the same).

Research approach

  1. The requests used the .request() method of NodeJS (v.14.18.1) of the “https” module with a timeout of 40 seconds and Chrome/96 version-level browser headers;
    code (very simplified)
    https.request({ hostname: host, port: 443, method: 'GET', headers: req_headers }, async (response) => {})
  2. All requests were made from servers located in the UK;
  3. All polls were made +-at night.

Additional Information

  1. observation – displays observations (by weeks +-) in dynamics from March 8 to May 8, 2022 (this information is also displayed on the chart, a couple of dates are omitted);
  2. comparison countries – for comparative analysis of the same key indicators with those of other countries;
  3. top sites – top 60 sites for clarity, with the display of one of the 4 states (availability or availability and direction of relocation).
  4. hosted zones – grouped domain zones of the 1st level, which represent the “original collection”;

Possible errors

  1. In general, for the indicator Unavailable there is an error that I did not determine, the machine polling method itself (in particular https.request()) does not imitate a regular User – some “erroneous” sites will be available when opened by a regular User through a browser.
    Factors that may erroneously affect resource unavailability
    • “protection from bots” (machine polling) could be connected;
    • ssl certificates might be outdated;
    • redirects or other restrictions may have been arranged.

    Therefore, “Unavailable” should not be identified with the complete unavailability of resources for Clients, the method was rather abandoned due to the simplicity of the mass check of resources in conditions of limited resources (more than 76 thousand resources, if we count for all the countries that I interviewed) and equal conditions with the initial period survey at the beginning of the year, where the same domains returned the correct result in 100% of cases.
    Also, a comparative analysis with Norway, where the indicator of “unavailable” resources is 1.5% with more resources (5.2k), suggests that the error in this regard should not be very high.

Some Conclusions

  • Comparative analysis shows:
    1. AT ~4.3 times “migration of resources” to other countries exceeded the average value;
    2. AT ~1.7 times “unavailability of resources” exceeded the average value;
  • In general, the relocation process slowed down, gaining slightly more than 1% over the past month, against 8% in the first month (since the beginning of observations);
  • 87% resources are located in the capital region (t={total resources}).

Disclaimer

Of course, hosting providers, search giants or large internet monitors will probably reflect the situation better, with more samples and better accessibility checks, I’m just sharing my observation, not claiming to be the ultimate truth;

PS

I myself watched the situation for a long time and probably would not have issued and published observations, but it was prompted by the fact that I could no longer renew the machine in the UK on European hosting (for well-known reasons for disabling cards), it would be a pity to lose the accumulated metric, so in the dynamics of the indicators are over, perhaps in the future I will summarize and post additionally (subscribe).

My contacts

Telegram: t.me/m_verhov (here is the original image and a version of the simplified-mobile version of the monitoring results)

linkedin: https://www.linkedin.com/in/verhov

twitter: https://twitter.com/michaelverhov

Can anyone share their observation, insider information or a situation related to the issue of forced relocation?

Similar Posts

Leave a Reply Cancel reply