1

TTL optimization set to One Week

Hi,

Can you specify a statement with the duration in cache optimization performance description  (300s if "boost" activated) in the setting tab ? It is not clear.

Is it possible to set TTL to One week or One day with a "super boost" option ?  it’s painful to sort through dozens of redundant and unnecessary requests every day, it’s not normal to have so much DNS traffic in normal daily operations.

Everybody knows how to flush the dns cache on their device or just rebooting their device if there's a major IP change into a parent DNS.

thanks

8 replies

null
    • Xavfree
    • 4 days ago
    • Reported - view

    One day would be a good start (with a warning statement if you want).

     

    https://www.tutoriaux-excalibur.com/threads/augmenter-la-taille-du-cache-dns-sur-iso-windows-11-25h2.41602/

    Sur Windows 11 25H2, le cache DNS est encore plus sollicite par les services de telemetrie IA et Microsoft Store.
    Par rapport a Windows 10, augmenter le MaxCacheTtl sur la 25H2 reduit significativement les micro-pauses de resolution lors de l'utilisation d'applications hybrides web/locales comme Discord ou Steam.

    Augmenter le MaxCacheTtl, c'est donner a ton Windows 25H2 un carnet d'adresses geant pour qu'il n'ait plus besoin de chercher son chemin a chaque carrefour

    • NextDNs
    • 4 days ago
    • Reported - view

    This would break many services.

      • Xavfree
      • 4 days ago
      • Reported - view

       I would really be curious to know which ones, and to be able to experience them for myself. Since Win10, the cache parameter is set by default to 86400s (1 day).

      Can you add this TTL switch parameter to One day for us, in a settings test mode only activable for 24H and one or two times in the month for example ?

      I'm always glad to see a succesful blocked query, but when I see this same query blocked hundred and hundred of times on the day this mean there's a problem somewhere.

      Do you not mind having hundreds and hundred of redundant lines and that this affects your global performance?

      • Xavfree
      • 2 days ago
      • Reported - view

       I assume you’re talking about services like "Apple" that constantly check network connectivity by making DNS requests, which is not a very clean way to work because it makes the DNS cache system useless.

      I don't think there are so much services working this way, a "lower TTL option" for those poorly implemented specific web services should do the job.

      Those TTL options would be great, and would give NextDNS users more possibilities to adjust according to their environment.

      • NextDNs
      • 4 hrs ago
      • Reported - view

       The Windows setting you mentioned (MaxCacheTtl) is not equivalent to increasing DNS TTLs. It only sets the maximum cache lifetime. Windows still respects shorter TTLs returned by DNS servers. For example, a record with a 30-second TTL will still expire after 30 seconds, even with MaxCacheTtl set to one week.

      Forcing a one-week TTL would override the service operator’s intended behavior and could cause real issues:

      • AWS Elastic Load Balancing uses 60-second TTLs to accommodate IP changes during scaling and failover.
      • Cloudflare Load Balancing uses 30-second TTLs for health-based traffic steering.
      • Akamai CDN uses TTLs as short as 20 seconds to dynamically direct users to healthy, optimal edge servers.
      • Route 53 / Azure Traffic Manager rely on short TTLs for regional failover and blue/green deployments.

      With a one-week TTL, clients could remain pinned to failed servers, retired deployments, or suboptimal endpoints for days.

      DNS TTLs aren’t just a caching optimization. They’re an essential part of how services manage availability and traffic. Overriding them globally would be unsafe.

      • Xavfree
      • 3 hrs ago
      • Reported - view

       Thank you for the complete answer 🙂

      I anderstand that windows can handle in the local cache a One day 86400s TTL query (max and by default) sent by DNS server.

      Those big companies load balancing services are working like proxies (very strange for private web browsing), the local server is identified and used. I anderstand one can be afraid with a One week TTL in an enterprise env. but One day TTL would not be a problem for a home user like me. Very critical load balancing services can be easily identified if there is a need to keep low level TTL for them.

      Would be great to test, because I anderstand that DNS cache system is in the way to be definitively useless... it's a turning point.

      It’s quite comical, since the traffic generated by these millions of DNS requests everywhere contradicts the desire to limit web traffic from these load balancing services.

      thanks for your interest.

      • NextDNs
      • 2 hrs ago
      • Reported - view

       This isn’t limited to a few poorly implemented services. Short TTLs are intentionally used by major infrastructure providers for load balancing, failover, CDN routing, and blue/green deployments.

      For home users, the impact of overriding these TTLs to one day or one week could be very noticeable. Websites could become unreachable, slow, or intermittently fail because your device keeps connecting to an IP address that is no longer valid or optimal. This isn’t something most users should have to troubleshoot by manually flushing their DNS cache.

      A 5-minute TTL doesn’t defeat the purpose of DNS caching either. The important distinction is between your device’s cache and the recursive resolver’s cache. When your device’s cache expires, NextDNS can typically answer directly from its own cache without performing a new lookup against authoritative DNS servers. We also proactively refresh cached records to minimize cache misses. The additional round trip to our resolver is generally negligible compared to a full recursive lookup.

      This is precisely where NextDNS Cache Boost helps. By increasing very short TTLs to a minimum of 5 minutes, it prevents applications and devices from repeatedly querying the resolver unnecessarily, while keeping the risk of stale DNS records reasonably limited. It strikes a balance between reducing DNS traffic and preserving the ability of services to adapt to infrastructure changes.

      Finally, repeated queries from applications that deliberately bypass caching won’t necessarily be reduced by increasing TTLs.

      The goal is to make DNS resolution fast and reliable, not simply to minimize the number of queries at the expense of correctness and availability.

      • Xavfree
      • 56 min ago
      • Reported - view

       Okay, thanks for clarifying here this "boost TTL" option. I anderstand now better your good work to maintain an efficient balance.

      I'm amazed to see the extent that the DNS systems have grown from internet start and also the disrespect of applications that mock them by bypassing constantly.

      I anderstand now why DNS servers are more and more implemented on home or pro routers. 

      thank you

Content aside

  • 1 Likes
  • 56 min agoLast active
  • 8Replies
  • 97Views
  • 2 Following