You are viewing a single comment's thread from:

RE: Crab Bucket - The STEEM Blockchain as a Local Database

in #steem • 8 years ago

I think it's great :).

I'm working on performance tuning for my golang implementation of this and similar ideas, and it does target ripping the whole chain. Tell me: Where did you find the limiting factor to be?

Your DB, or STEEMD?

I think it's steemd, but I'm not 100000% certain, so I'm curious what your experience has been so far.

8 years ago in #steem by
$0.00
    Reply 2
    Sort:  
  • Trending
    • [-]
     8 years ago 

    I think my db is a bottleneck. I was only testing with sqlite. I can easily switch to postgres and see. But I have a feeling that steemd presents its own bottleneck. It's fine if you're streaming because the blocks come in once every 3 seconds. But replay/rebuild takes 1 second per block over node.steem.ws (not recommended), which is pretty dang slow.

    I haven't gotten my local node to work quite right, so I haven't seen the perfect scenario yet.

    $0.00
      Reply
      [-]
       8 years ago 

      oh, I use a local node. Do you you want a hand getting that straight?

      For me I top out at about 500 blocks per second and average about 350-400. Nonetheless, takes a solid 24 hours to do the dump.

      $0.00
        Reply