• Tja@programming.dev
    link
    fedilink
    arrow-up
    1
    ·
    4 months ago

    Azure has had nvme drives with hundreds of thousands of iops for a couple of years. That plus sharding should take care of any problems. I have personally taken care of a project with 2M tps and it wasn’t even the biggest in my region, at the small-ish company I work for. Yes, re-architecting that is a pain but doable.

    Lift and shift may be out of the question tho.

    • Jo Miran@lemmy.ml
      link
      fedilink
      arrow-up
      1
      ·
      3 months ago

      Lift and shift may be out of the question tho.

      That and cost are the real issue. This is not their app, so they cannot rearchitect it. To make it fit to best use cloud resources, the thing would need to be rebuilt from scratch, and the savings are just not there to do so.

      We need to move 8 gigabytes per second, sustained, which Azure can do at their highest tier. The problem is that, as I said, this is the smallest of our medium clients, and at the tier that can sustain that, the cost per terabyte per month is well beyond the client’s budget. We told them to go with Exadata but the initial sticker shock and the execs lack of faith on their IT team made them run. Now they are committed to Azure. Just last week we had that call where we told Azure our requirements and they just came back with the price a few days ago. Suddenly, dropping seven figures for the hardware doesn’t seem too expensive.

      Whatever. We get paid either way.