Record🤔

uri:

"at://did:plc:hdhoaan3xa3jiuq4fg4mefid/app.bsky.feed.post/3ld2zn3fi6c2u"

cid:

"bafyreibqihk36nz452ieaipvlvchcmrfz2swnsql2r4wajy7grb6ysc2ky"

value:

text:

"my fave db chart yet: kv random read performance! both axes are on log scales here. some surprises!

redb is absurdly fast but the btrees (so sqlite too) have write performance limits.

i did NOT expect rocks to outperform sqlite here!? significantly!!

fjall puts in an impressive effort (cont’d)"

$type:

"app.bsky.feed.post"

embed:

$type:

"app.bsky.embed.recordWithMedia"

media:

$type:

"app.bsky.embed.images"

images:

• alt:

  "scatter plot of key lookup time vs value size for redb, rocks, sqlite, and fjall. the results span orders of magnitude: the slowest redb time is close to fjall’s fastest time!
  
  the shapes from redb, rocks, and sqlite are about the same but with big performance gaps (fastest to slowest)
  
  fjall is the worst for small values, but holds consistent performance past where the other dbs’ performance curves start to bend, allowing it to overtake sqlite for speed with larger values."

  image:

  

  View blob content

  $type:

  "blob"

  ref:

  $link:

  bafkreiblft6e2bpe7lxdrhjtre535wosexhvcsb3herdvv62vow3ayvq2y

  mimeType:

  "image/jpeg"

  size:

  332156

  aspectRatio:

  width:

  1462

  height:

  1008

record:

$type:

"app.bsky.embed.record"

record:

cid:

"bafyreiai6csy47fvuiwsu5aswdc4z7eqjuskjxgeitu6uqesbthtxyivjq"

uri:

"at://did:plc:hdhoaan3xa3jiuq4fg4mefid/app.bsky.feed.post/3ld22elodlk23"

langs:

• "en"

createdAt:

"2024-12-12T00:36:01.951Z"