Mark Callaghan

@MarkCallaghanDB

Database efficiency & performance. I work at . Opinions are my own -

Vrijeme pridruživanja: rujan 2019.

Tweetovi

Blokirali ste korisnika/cu @MarkCallaghanDB

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @MarkCallaghanDB

  1. My question is whether the added durability and availability from 6 copies is worth the cost. Where that cost is measured in power, space, dollars. But it is hard to estimate the benefit from the outside. Fewer copies is good for the environment.

    Prikaži ovu nit
    Poništi
  2. The Aurora paper made a great argument on why 6 copies are used. But I am curious whether this was quantified and whether some customers would prefer fewer copies if that reduced their bill.

    Prikaži ovu nit
    Poništi
  3. For #2 what is the return in terms of durability and availability when going from 1 to 6 copies? I have no doubt that 6 is better. My question is about the return on investment. This isn’t a good advertisement for the cloud if 6 copies are required to make things work.

    Prikaži ovu nit
    Poništi
  4. For #1 it is obvious that Aurora is ahead, but is the revenue difference closer to 10X or 2X? AWS uses a pragmatic approach — build on MySQL — while Google uses a Google approach. There is a huge difference per

    Prikaži ovu nit
    Poništi
  5. Cloud database questions: 1) How is Cloud Spanner doing versus AWS Aurora? 2) Does AWS Aurora need 6 copies of the database?

    Prikaži ovu nit
    Poništi
  6. 31. sij

    Hope I didn't make too many mistakes. I am not a distributed systems guru.

    Poništi
  7. 31. sij
    Poništi
  8. 31. sij

    subscribed so that I don't miss these again. Content is excellent. XKCD references are a bonus.

    Prikaži ovu nit
    Poništi
  9. 31. sij
    Prikaži ovu nit
    Poništi
  10. 30. sij

    Is there an easier way to list files installed for an apt package? sudo apt install apt-file sudo apt-file update apt-file list $package

    Poništi
  11. 30. sij

    To see tcmalloc size classes in serverStatus output with MongoDB use tcmalloc:2 db.serverStatus({tcmalloc:2}).tcmalloc For basic tcmalloc stats use nothing or tcmalloc:1 db.serverStatus().tcmalloc db.serverStatus({tcmalloc:1}).tcmalloc

    Poništi
  12. 30. sij
    Prikaži ovu nit
    Poništi
  13. 29. sij
    Prikaži ovu nit
    Poništi
  14. 29. sij

    Then aggregate it for one thread: cat all.f | awk 'BEGIN { s = ""; } /^Thread/ { print s; s = ""; } /^\#/ { x=index($2, "0x"); if (x == 1) { n=$4 } else { n=$2 }; if (s != "" ) { s = s "," n} else { s = n } } END { print s }' - | sort | uniq -c | sort -r -n -k 1,1

    Prikaži ovu nit
    Poništi
  15. 29. sij

    First time I tried this. Get all call stacks for one thread from N PMP runs: for d in f.*; do echo ; awk '/\(LWP 23839/,/^$/' ; done > all.f

    Prikaži ovu nit
    Poništi
  16. 28. sij
    Poništi
  17. 28. sij

    I was reminded of XFS speculative allocation thanks to buffered IO.

    Poništi
  18. 27. sij

    So I am thrilled to learn that Manos Athanassoulis and team are working on this -- They also have work on bloom filters - Too bad I can't be at NEDB this year.

    Prikaži ovu nit
    Poništi
  19. 27. sij

    Optimizing for tombstone removal in an LSM is a hard problem

    Prikaži ovu nit
    Poništi
  20. 25. sij

    I am happy to see companies like leverage so they can focus on adding value higher in the stack. Just like FB was able to leverage LevelDB to start the RocksDB project -

    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·