Run TPC-C on Pure and on SQLite, and compare the states they leave

1 files changed, 68 insertions, 0 deletions

diff --git a/benchmarks/tpcc/data.py b/benchmarks/tpcc/data.py
new file mode 100644
index 0000000..d8a4ccb
--- /dev/null
+++ b/benchmarks/tpcc/data.py
@@ -0,0 +1,68 @@
+"""TPC-C's initial database (clause 4.3.3.1) for W warehouses, as rows of the
+records in tpcc.aup (each composite key packed into one number, as there),
+and the same rows for SQLite's tables with their own keys. Texts the
+transactions only carry along are left out (see tpcc.aup)."""
+import random
+import struct
+
+ITEMS, CUSTOMERS, ORDERS, DELIVERED = 100000, 3000, 3000, 2100
+
+# Each table's fields in record order (tpcc.aup), all numbers.
+FIELDS = {
+    'warehouses': ('id', 'tax', 'ytd'),
+    'districts': ('id', 'tax', 'ytd', 'next_order', 'next_delivery'),
+    'customers': ('id', 'discount', 'credit', 'balance', 'ytd_payment', 'payments', 'deliveries', 'last_order'),
+    'items': ('id', 'price'),
+    'stock': ('id', 'quantity', 'ytd', 'orders', 'remote'),
+    'orders': ('id', 'customer', 'entry', 'carrier', 'lines', 'local'),
+    'new_orders': ('id',),
+    'lines': ('id', 'item', 'supply', 'delivery', 'quantity', 'amount'),
+}
+
+
+def district_key(w, d):
+    return w * 10 + d
+
+
+def customer_key(w, d, c):
+    return district_key(w, d) * 3000 + c
+
+
+def order_key(w, d, o):
+    return district_key(w, d) * 10000000 + o
+
+
+def generate(warehouses, seed=1, now=1.7e12):
+    """The tables as lists of tuples in FIELDS order."""
+    rng = random.Random(seed)
+    tables = {name: [] for name in FIELDS}
+    tables['items'] = [(i, round(rng.uniform(1, 100), 2)) for i in range(1, ITEMS + 1)]
+    for w in range(1, warehouses + 1):
+        tables['warehouses'].append((w, round(rng.uniform(0, 0.2), 4), 300000.0))
+        tables['stock'].extend((w * 100000 + i, rng.randint(10, 100), 0, 0, 0) for i in range(1, ITEMS + 1))
+        for d in range(1, 11):
+            tables['districts'].append((district_key(w, d), round(rng.uniform(0, 0.2), 4), 30000.0, ORDERS + 1,
+                                        DELIVERED + 1))
+            owners = list(range(1, CUSTOMERS + 1))
+            rng.shuffle(owners)
+            last = {}
+            for o in range(1, ORDERS + 1):
+                key, count, delivered = order_key(w, d, o), rng.randint(5, 15), o <= DELIVERED
+                tables['orders'].append((key, customer_key(w, d, owners[o - 1]), now,
+                                         rng.randint(1, 10) if delivered else 0, count, 1))
+                last[owners[o - 1]] = key
+                if not delivered:
+                    tables['new_orders'].append((key,))
+                for n in range(1, count + 1):
+                    tables['lines'].append((key * 16 + n, rng.randint(1, ITEMS), w, now if delivered else 0, 5,
+                                            0 if delivered else round(rng.uniform(0.01, 9999.99), 2)))
+            for c in range(1, CUSTOMERS + 1):
+                tables['customers'].append((customer_key(w, d, c), round(rng.uniform(0, 0.5), 4),
+                                            0 if rng.random() < 0.1 else 1, -10.0, 10.0, 1, 0, last[c]))
+    return tables
+
+
+def packed(rows):
+    """Rows in a list's stored form (each number's binary64, little-endian)."""
+    width = len(rows[0])
+    return struct.pack(f'<{width * len(rows)}d', *(v for row in rows for v in row))