|
| 1 | +use std::{collections::HashSet, sync::Arc, time::Instant}; |
| 2 | + |
| 3 | +use anyhow::anyhow; |
| 4 | +use clap::Parser; |
| 5 | +use graph::{ |
| 6 | + components::store::write::{EntityModification, RowGroupForPerfTest as RowGroup}, |
| 7 | + data::{ |
| 8 | + store::{Id, Value}, |
| 9 | + subgraph::DeploymentHash, |
| 10 | + value::Word, |
| 11 | + }, |
| 12 | + schema::{EntityType, InputSchema}, |
| 13 | +}; |
| 14 | +use lazy_static::lazy_static; |
| 15 | +use rand::{rng, Rng}; |
| 16 | + |
| 17 | +#[derive(Parser)] |
| 18 | +#[clap( |
| 19 | + name = "append_row", |
| 20 | + about = "Measure time it takes to append rows to a row group" |
| 21 | +)] |
| 22 | +struct Opt { |
| 23 | + /// Number of repetitions of the test |
| 24 | + #[clap(short, long, default_value = "5")] |
| 25 | + niter: usize, |
| 26 | + /// Number of rows |
| 27 | + #[clap(short, long, default_value = "10000")] |
| 28 | + rows: usize, |
| 29 | + /// Number of blocks |
| 30 | + #[clap(short, long, default_value = "300")] |
| 31 | + blocks: usize, |
| 32 | + /// Number of ids |
| 33 | + #[clap(short, long, default_value = "500")] |
| 34 | + ids: usize, |
| 35 | +} |
| 36 | + |
| 37 | +// A very fake schema that allows us to get the entity types we need |
| 38 | +const GQL: &str = r#" |
| 39 | + type Thing @entity { id: ID!, count: Int! } |
| 40 | + type RowGroup @entity { id: ID! } |
| 41 | + type Entry @entity { id: ID! } |
| 42 | + "#; |
| 43 | +lazy_static! { |
| 44 | + static ref DEPLOYMENT: DeploymentHash = DeploymentHash::new("batchAppend").unwrap(); |
| 45 | + static ref SCHEMA: InputSchema = InputSchema::parse_latest(GQL, DEPLOYMENT.clone()).unwrap(); |
| 46 | + static ref THING_TYPE: EntityType = SCHEMA.entity_type("Thing").unwrap(); |
| 47 | + static ref ROW_GROUP_TYPE: EntityType = SCHEMA.entity_type("RowGroup").unwrap(); |
| 48 | + static ref ENTRY_TYPE: EntityType = SCHEMA.entity_type("Entry").unwrap(); |
| 49 | +} |
| 50 | + |
| 51 | +pub fn main() -> anyhow::Result<()> { |
| 52 | + let opt = Opt::parse(); |
| 53 | + let next_block = opt.blocks as f64 / opt.rows as f64; |
| 54 | + for _ in 0..opt.niter { |
| 55 | + let ids = (0..opt.ids) |
| 56 | + .map(|n| Id::String(Word::from(format!("00{n}010203040506")))) |
| 57 | + .collect::<Vec<_>>(); |
| 58 | + let mut existing: HashSet<Id> = HashSet::new(); |
| 59 | + let mut mods = Vec::new(); |
| 60 | + let mut block = 0; |
| 61 | + let mut block_pos = Vec::new(); |
| 62 | + for _ in 0..opt.rows { |
| 63 | + if rng().random_bool(next_block) { |
| 64 | + block += 1; |
| 65 | + block_pos.clear(); |
| 66 | + } |
| 67 | + |
| 68 | + let mut attempt = 0; |
| 69 | + let pos = loop { |
| 70 | + if attempt > 20 { |
| 71 | + return Err(anyhow!( |
| 72 | + "Failed to find a position in 20 attempts. Increase `ids`" |
| 73 | + )); |
| 74 | + } |
| 75 | + attempt += 1; |
| 76 | + let pos = rng().random_range(0..opt.ids); |
| 77 | + if block_pos.contains(&pos) { |
| 78 | + continue; |
| 79 | + } |
| 80 | + block_pos.push(pos); |
| 81 | + break pos; |
| 82 | + }; |
| 83 | + let id = &ids[pos]; |
| 84 | + let data = vec![ |
| 85 | + (Word::from("id"), Value::String(id.to_string())), |
| 86 | + (Word::from("count"), Value::Int(block as i32)), |
| 87 | + ]; |
| 88 | + let data = Arc::new(SCHEMA.make_entity(data).unwrap()); |
| 89 | + let md = if existing.contains(id) { |
| 90 | + EntityModification::Overwrite { |
| 91 | + key: THING_TYPE.key(id.clone()), |
| 92 | + data, |
| 93 | + block, |
| 94 | + end: None, |
| 95 | + } |
| 96 | + } else { |
| 97 | + existing.insert(id.clone()); |
| 98 | + EntityModification::Insert { |
| 99 | + key: THING_TYPE.key(id.clone()), |
| 100 | + data, |
| 101 | + block, |
| 102 | + end: None, |
| 103 | + } |
| 104 | + }; |
| 105 | + mods.push(md); |
| 106 | + } |
| 107 | + let mut group = RowGroup::new(THING_TYPE.clone(), false); |
| 108 | + |
| 109 | + let start = Instant::now(); |
| 110 | + for md in mods { |
| 111 | + group.append_row(md).unwrap(); |
| 112 | + } |
| 113 | + let elapsed = start.elapsed(); |
| 114 | + println!( |
| 115 | + "Adding {} rows with {} ids across {} blocks took {:?}", |
| 116 | + opt.rows, |
| 117 | + existing.len(), |
| 118 | + block, |
| 119 | + elapsed |
| 120 | + ); |
| 121 | + } |
| 122 | + Ok(()) |
| 123 | +} |
0 commit comments