Concepte de baza in lumea cloudului. HA si DR

แชร์
ฝัง
  • เผยแพร่เมื่อ 10 พ.ย. 2024

ความคิดเห็น • 4

  • @return_1101
    @return_1101 2 ปีที่แล้ว

    Interesant. Continuă. Aflăm chestii foarte bune.

  • @claudiu1984c
    @claudiu1984c ปีที่แล้ว

    Din cate cunosc chiar si cu MySQL HA enabed, aplicatia tot va avea downtime, cel putin in AWS RDS in caz de fail pe master, dureaza cam 1 minut failoverul.
    Iar la DR, oh well... E mult mai complex.

    • @RAVTechz
      @RAVTechz  ปีที่แล้ว

      Salut, da. Pentru ca Mysql are replicare async by default. Daca vrei replicare sincrona cu DR si downtime cat mai mic se poate apela la o tehnologie terta precum Mysql Percona xtradb si practic masterul ar fi un cluster de 3 masini iar replicile la fel. Daca iti pica un server de master mai ai 2 up and running. Ce vreau sa spun este ca exista niste arhitecruri mai complexe pentru a minimiza downtime-ul.

    • @claudiu1984c
      @claudiu1984c ปีที่แล้ว

      ​@@RAVTechz Probabil avem o definitie diferita pt DR. In termenii folositi de mine, DR implica cross-region replication. PXC ofera HA(same region replication) cu sync replication. DR adevarat(cross-region), nu ar putea functiona cu sync replication, ar fi delay mare la scriere pe master.
      Noi avem DR implementat in AWS RDS prin async cross-region replication iar in K8S folosim PXC unde facem CR repl. la K8s volumes folosind Portworx.
      Cheers.