Replace uses of simple_insert_many with simple_insert_many_values. (#11742)

This should be (slightly) more efficient and it is simpler
to have a single method for inserting multiple values.

19 files changed, 263 insertions, 298 deletions

diff --git a/changelog.d/11742.misc b/changelog.d/11742.misc
new file mode 100644
index 0000000000..f65ccdf30a
--- /dev/null
+++ b/changelog.d/11742.misc
@@ -0,0 +1 @@
+Minor efficiency improvements when inserting many values into the database.
diff --git a/synapse/rest/admin/background_updates.py b/synapse/rest/admin/background_updates.py
index 6ec00ce0b9..e9bce22a34 100644
--- a/synapse/rest/admin/background_updates.py
+++ b/synapse/rest/admin/background_updates.py
@@ -123,34 +123,25 @@ class BackgroundUpdateStartJobRestServlet(RestServlet):
         job_name = body["job_name"]
 
         if job_name == "populate_stats_process_rooms":
-            jobs = [
-                {
-                    "update_name": "populate_stats_process_rooms",
-                    "progress_json": "{}",
-                },
-            ]
+            jobs = [("populate_stats_process_rooms", "{}", "")]
         elif job_name == "regenerate_directory":
             jobs = [
-                {
-                    "update_name": "populate_user_directory_createtables",
-                    "progress_json": "{}",
-                    "depends_on": "",
-                },
-                {
-                    "update_name": "populate_user_directory_process_rooms",
-                    "progress_json": "{}",
-                    "depends_on": "populate_user_directory_createtables",
-                },
-                {
-                    "update_name": "populate_user_directory_process_users",
-                    "progress_json": "{}",
-                    "depends_on": "populate_user_directory_process_rooms",
-                },
-                {
-                    "update_name": "populate_user_directory_cleanup",
-                    "progress_json": "{}",
-                    "depends_on": "populate_user_directory_process_users",
-                },
+                ("populate_user_directory_createtables", "{}", ""),
+                (
+                    "populate_user_directory_process_rooms",
+                    "{}",
+                    "populate_user_directory_createtables",
+                ),
+                (
+                    "populate_user_directory_process_users",
+                    "{}",
+                    "populate_user_directory_process_rooms",
+                ),
+                (
+                    "populate_user_directory_cleanup",
+                    "{}",
+                    "populate_user_directory_process_users",
+                ),
             ]
         else:
             raise SynapseError(HTTPStatus.BAD_REQUEST, "Invalid job_name")
@@ -158,6 +149,7 @@ class BackgroundUpdateStartJobRestServlet(RestServlet):
         try:
             await self._store.db_pool.simple_insert_many(
                 table="background_updates",
+                keys=("update_name", "progress_json", "depends_on"),
                 values=jobs,
                 desc=f"admin_api_run_{job_name}",
             )
diff --git a/synapse/storage/database.py b/synapse/storage/database.py
index a27cc3605c..57cc1d76e0 100644
--- a/synapse/storage/database.py
+++ b/synapse/storage/database.py
@@ -934,56 +934,6 @@ class DatabasePool:
         txn.execute(sql, vals)
 
     async def simple_insert_many(
-        self, table: str, values: List[Dict[str, Any]], desc: str
-    ) -> None:
-        """Executes an INSERT query on the named table.
-
-        The input is given as a list of dicts, with one dict per row.
-        Generally simple_insert_many_values should be preferred for new code.
-
-        Args:
-            table: string giving the table name
-            values: dict of new column names and values for them
-            desc: description of the transaction, for logging and metrics
-        """
-        await self.runInteraction(desc, self.simple_insert_many_txn, table, values)
-
-    @staticmethod
-    def simple_insert_many_txn(
-        txn: LoggingTransaction, table: str, values: List[Dict[str, Any]]
-    ) -> None:
-        """Executes an INSERT query on the named table.
-
-        The input is given as a list of dicts, with one dict per row.
-        Generally simple_insert_many_values_txn should be preferred for new code.
-
-        Args:
-            txn: The transaction to use.
-            table: string giving the table name
-            values: dict of new column names and values for them
-        """
-        if not values:
-            return
-
-        # This is a *slight* abomination to get a list of tuples of key names
-        # and a list of tuples of value names.
-        #
-        # i.e. [{"a": 1, "b": 2}, {"c": 3, "d": 4}]
-        #         => [("a", "b",), ("c", "d",)] and [(1, 2,), (3, 4,)]
-        #
-        # The sort is to ensure that we don't rely on dictionary iteration
-        # order.
-        keys, vals = zip(
-            *(zip(*(sorted(i.items(), key=lambda kv: kv[0]))) for i in values if i)
-        )
-
-        for k in keys:
-            if k != keys[0]:
-                raise RuntimeError("All items must have the same keys")
-
-        return DatabasePool.simple_insert_many_values_txn(txn, table, keys[0], vals)
-
-    async def simple_insert_many_values(
         self,
         table: str,
         keys: Collection[str],
@@ -1002,11 +952,11 @@ class DatabasePool:
             desc: description of the transaction, for logging and metrics
         """
         await self.runInteraction(
-            desc, self.simple_insert_many_values_txn, table, keys, values
+            desc, self.simple_insert_many_txn, table, keys, values
         )
 
     @staticmethod
-    def simple_insert_many_values_txn(
+    def simple_insert_many_txn(
         txn: LoggingTransaction,
         table: str,
         keys: Collection[str],
diff --git a/synapse/storage/databases/main/account_data.py b/synapse/storage/databases/main/account_data.py
index 93db71d1b4..ef475e18c7 100644
--- a/synapse/storage/databases/main/account_data.py
+++ b/synapse/storage/databases/main/account_data.py
@@ -536,9 +536,9 @@ class AccountDataWorkerStore(CacheInvalidationWorkerStore):
         self.db_pool.simple_insert_many_txn(
             txn,
             table="ignored_users",
+            keys=("ignorer_user_id", "ignored_user_id"),
             values=[
-                {"ignorer_user_id": user_id, "ignored_user_id": u}
-                for u in currently_ignored_users - previously_ignored_users
+                (user_id, u) for u in currently_ignored_users - previously_ignored_users
             ],
         )
 
diff --git a/synapse/storage/databases/main/deviceinbox.py b/synapse/storage/databases/main/deviceinbox.py
index 3682cb6a81..4eca97189b 100644
--- a/synapse/storage/databases/main/deviceinbox.py
+++ b/synapse/storage/databases/main/deviceinbox.py
@@ -432,14 +432,21 @@ class DeviceInboxWorkerStore(SQLBaseStore):
             self.db_pool.simple_insert_many_txn(
                 txn,
                 table="device_federation_outbox",
+                keys=(
+                    "destination",
+                    "stream_id",
+                    "queued_ts",
+                    "messages_json",
+                    "instance_name",
+                ),
                 values=[
-                    {
-                        "destination": destination,
-                        "stream_id": stream_id,
-                        "queued_ts": now_ms,
-                        "messages_json": json_encoder.encode(edu),
-                        "instance_name": self._instance_name,
-                    }
+                    (
+                        destination,
+                        stream_id,
+                        now_ms,
+                        json_encoder.encode(edu),
+                        self._instance_name,
+                    )
                     for destination, edu in remote_messages_by_destination.items()
                 ],
             )
@@ -571,14 +578,9 @@ class DeviceInboxWorkerStore(SQLBaseStore):
         self.db_pool.simple_insert_many_txn(
             txn,
             table="device_inbox",
+            keys=("user_id", "device_id", "stream_id", "message_json", "instance_name"),
             values=[
-                {
-                    "user_id": user_id,
-                    "device_id": device_id,
-                    "stream_id": stream_id,
-                    "message_json": message_json,
-                    "instance_name": self._instance_name,
-                }
+                (user_id, device_id, stream_id, message_json, self._instance_name)
                 for user_id, messages_by_device in local_by_user_then_device.items()
                 for device_id, message_json in messages_by_device.items()
             ],
diff --git a/synapse/storage/databases/main/devices.py b/synapse/storage/databases/main/devices.py
index 52fbf50db6..8748654b55 100644
--- a/synapse/storage/databases/main/devices.py
+++ b/synapse/storage/databases/main/devices.py
@@ -1386,12 +1386,9 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore):
         self.db_pool.simple_insert_many_txn(
             txn,
             table="device_lists_remote_cache",
+            keys=("user_id", "device_id", "content"),
             values=[
-                {
-                    "user_id": user_id,
-                    "device_id": content["device_id"],
-                    "content": json_encoder.encode(content),
-                }
+                (user_id, content["device_id"], json_encoder.encode(content))
                 for content in devices
             ],
         )
@@ -1479,8 +1476,9 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore):
         self.db_pool.simple_insert_many_txn(
             txn,
             table="device_lists_stream",
+            keys=("stream_id", "user_id", "device_id"),
             values=[
-                {"stream_id": stream_id, "user_id": user_id, "device_id": device_id}
+                (stream_id, user_id, device_id)
                 for stream_id, device_id in zip(stream_ids, device_ids)
             ],
         )
@@ -1507,18 +1505,27 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore):
         self.db_pool.simple_insert_many_txn(
             txn,
             table="device_lists_outbound_pokes",
+            keys=(
+                "destination",
+                "stream_id",
+                "user_id",
+                "device_id",
+                "sent",
+                "ts",
+                "opentracing_context",
+            ),
             values=[
-                {
-                    "destination": destination,
-                    "stream_id": next(next_stream_id),
-                    "user_id": user_id,
-                    "device_id": device_id,
-                    "sent": False,
-                    "ts": now,
-                    "opentracing_context": json_encoder.encode(context)
+                (
+                    destination,
+                    next(next_stream_id),
+                    user_id,
+                    device_id,
+                    False,
+                    now,
+                    json_encoder.encode(context)
                     if whitelisted_homeserver(destination)
                     else "{}",
-                }
+                )
                 for destination in hosts
                 for device_id in device_ids
             ],
diff --git a/synapse/storage/databases/main/directory.py b/synapse/storage/databases/main/directory.py
index f76c6121e8..5903fdaf00 100644
--- a/synapse/storage/databases/main/directory.py
+++ b/synapse/storage/databases/main/directory.py
@@ -112,10 +112,8 @@ class DirectoryWorkerStore(CacheInvalidationWorkerStore):
             self.db_pool.simple_insert_many_txn(
                 txn,
                 table="room_alias_servers",
-                values=[
-                    {"room_alias": room_alias.to_string(), "server": server}
-                    for server in servers
-                ],
+                keys=("room_alias", "server"),
+                values=[(room_alias.to_string(), server) for server in servers],
             )
 
             self._invalidate_cache_and_stream(
diff --git a/synapse/storage/databases/main/e2e_room_keys.py b/synapse/storage/databases/main/e2e_room_keys.py
index 0cb48b9dd7..b789a588a5 100644
--- a/synapse/storage/databases/main/e2e_room_keys.py
+++ b/synapse/storage/databases/main/e2e_room_keys.py
@@ -110,16 +110,16 @@ class EndToEndRoomKeyStore(SQLBaseStore):
         values = []
         for (room_id, session_id, room_key) in room_keys:
             values.append(
-                {
-                    "user_id": user_id,
-                    "version": version_int,
-                    "room_id": room_id,
-                    "session_id": session_id,
-                    "first_message_index": room_key["first_message_index"],
-                    "forwarded_count": room_key["forwarded_count"],
-                    "is_verified": room_key["is_verified"],
-                    "session_data": json_encoder.encode(room_key["session_data"]),
-                }
+                (
+                    user_id,
+                    version_int,
+                    room_id,
+                    session_id,
+                    room_key["first_message_index"],
+                    room_key["forwarded_count"],
+                    room_key["is_verified"],
+                    json_encoder.encode(room_key["session_data"]),
+                )
             )
             log_kv(
                 {
@@ -131,7 +131,19 @@ class EndToEndRoomKeyStore(SQLBaseStore):
             )
 
         await self.db_pool.simple_insert_many(
-            table="e2e_room_keys", values=values, desc="add_e2e_room_keys"
+            table="e2e_room_keys",
+            keys=(
+                "user_id",
+                "version",
+                "room_id",
+                "session_id",
+                "first_message_index",
+                "forwarded_count",
+                "is_verified",
+                "session_data",
+            ),
+            values=values,
+            desc="add_e2e_room_keys",
         )
 
     @trace
diff --git a/synapse/storage/databases/main/end_to_end_keys.py b/synapse/storage/databases/main/end_to_end_keys.py
index 86cab97563..1f8447b507 100644
--- a/synapse/storage/databases/main/end_to_end_keys.py
+++ b/synapse/storage/databases/main/end_to_end_keys.py
@@ -387,15 +387,16 @@ class EndToEndKeyWorkerStore(EndToEndKeyBackgroundStore, CacheInvalidationWorker
             self.db_pool.simple_insert_many_txn(
                 txn,
                 table="e2e_one_time_keys_json",
+                keys=(
+                    "user_id",
+                    "device_id",
+                    "algorithm",
+                    "key_id",
+                    "ts_added_ms",
+                    "key_json",
+                ),
                 values=[
-                    {
-                        "user_id": user_id,
-                        "device_id": device_id,
-                        "algorithm": algorithm,
-                        "key_id": key_id,
-                        "ts_added_ms": time_now,
-                        "key_json": json_bytes,
-                    }
+                    (user_id, device_id, algorithm, key_id, time_now, json_bytes)
                     for algorithm, key_id, json_bytes in new_keys
                 ],
             )
@@ -1186,15 +1187,22 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore):
         """
         await self.db_pool.simple_insert_many(
             "e2e_cross_signing_signatures",
-            [
-                {
-                    "user_id": user_id,
-                    "key_id": item.signing_key_id,
-                    "target_user_id": item.target_user_id,
-                    "target_device_id": item.target_device_id,
-                    "signature": item.signature,
-                }
+            keys=(
+                "user_id",
+                "key_id",
+                "target_user_id",
+                "target_device_id",
+                "signature",
+            ),
+            values=[
+                (
+                    user_id,
+                    item.signing_key_id,
+                    item.target_user_id,
+                    item.target_device_id,
+                    item.signature,
+                )
                 for item in signatures
             ],
-            "add_e2e_signing_key",
+            desc="add_e2e_signing_key",
         )
diff --git a/synapse/storage/databases/main/event_push_actions.py b/synapse/storage/databases/main/event_push_actions.py
index a98e6b2593..b7c4c62222 100644
--- a/synapse/storage/databases/main/event_push_actions.py
+++ b/synapse/storage/databases/main/event_push_actions.py
@@ -875,14 +875,21 @@ class EventPushActionsWorkerStore(SQLBaseStore):
         self.db_pool.simple_insert_many_txn(
             txn,
             table="event_push_summary",
+            keys=(
+                "user_id",
+                "room_id",
+                "notif_count",
+                "unread_count",
+                "stream_ordering",
+            ),
             values=[
-                {
-                    "user_id": user_id,
-                    "room_id": room_id,
-                    "notif_count": summary.notif_count,
-                    "unread_count": summary.unread_count,
-                    "stream_ordering": summary.stream_ordering,
-                }
+                (
+                    user_id,
+                    room_id,
+                    summary.notif_count,
+                    summary.unread_count,
+                    summary.stream_ordering,
+                )
                 for ((user_id, room_id), summary) in summaries.items()
                 if summary.old_user_id is None
             ],
diff --git a/synapse/storage/databases/main/events.py b/synapse/storage/databases/main/events.py
index cce2305597..de3b48524b 100644
--- a/synapse/storage/databases/main/events.py
+++ b/synapse/storage/databases/main/events.py
@@ -442,12 +442,9 @@ class PersistEventsStore:
         self.db_pool.simple_insert_many_txn(
             txn,
             table="event_auth",
+            keys=("event_id", "room_id", "auth_id"),
             values=[
-                {
-                    "event_id": event.event_id,
-                    "room_id": event.room_id,
-                    "auth_id": auth_id,
-                }
+                (event.event_id, event.room_id, auth_id)
                 for event in events
                 for auth_id in event.auth_event_ids()
                 if event.is_state()
@@ -675,8 +672,9 @@ class PersistEventsStore:
         db_pool.simple_insert_many_txn(
             txn,
             table="event_auth_chains",
+            keys=("event_id", "chain_id", "sequence_number"),
             values=[
-                {"event_id": event_id, "chain_id": c_id, "sequence_number": seq}
+                (event_id, c_id, seq)
                 for event_id, (c_id, seq) in new_chain_tuples.items()
             ],
         )
@@ -782,13 +780,14 @@ class PersistEventsStore:
         db_pool.simple_insert_many_txn(
             txn,
             table="event_auth_chain_links",
+            keys=(
+                "origin_chain_id",
+                "origin_sequence_number",
+                "target_chain_id",
+                "target_sequence_number",
+            ),
             values=[
-                {
-                    "origin_chain_id": source_id,
-                    "origin_sequence_number": source_seq,
-                    "target_chain_id": target_id,
-                    "target_sequence_number": target_seq,
-                }
+                (source_id, source_seq, target_id, target_seq)
                 for (
                     source_id,
                     source_seq,
@@ -943,20 +942,28 @@ class PersistEventsStore:
             txn_id = getattr(event.internal_metadata, "txn_id", None)
             if token_id and txn_id:
                 to_insert.append(
-                    {
-                        "event_id": event.event_id,
-                        "room_id": event.room_id,
-                        "user_id": event.sender,
-                        "token_id": token_id,
-                        "txn_id": txn_id,
-                        "inserted_ts": self._clock.time_msec(),
-                    }
+                    (
+                        event.event_id,
+                        event.room_id,
+                        event.sender,
+                        token_id,
+                        txn_id,
+                        self._clock.time_msec(),
+                    )
                 )
 
         if to_insert:
             self.db_pool.simple_insert_many_txn(
                 txn,
                 table="event_txn_id",
+                keys=(
+                    "event_id",
+                    "room_id",
+                    "user_id",
+                    "token_id",
+                    "txn_id",
+                    "inserted_ts",
+                ),
                 values=to_insert,
             )
 
@@ -1161,8 +1168,9 @@ class PersistEventsStore:
         self.db_pool.simple_insert_many_txn(
             txn,
             table="event_forward_extremities",
+            keys=("event_id", "room_id"),
             values=[
-                {"event_id": ev_id, "room_id": room_id}
+                (ev_id, room_id)
                 for room_id, new_extrem in new_forward_extremities.items()
                 for ev_id in new_extrem
             ],
@@ -1174,12 +1182,9 @@ class PersistEventsStore:
         self.db_pool.simple_insert_many_txn(
             txn,
             table="stream_ordering_to_exterm",
+            keys=("room_id", "event_id", "stream_ordering"),
             values=[
-                {
-                    "room_id": room_id,
-                    "event_id": event_id,
-                    "stream_ordering": max_stream_order,
-                }
+                (room_id, event_id, max_stream_order)
                 for room_id, new_extrem in new_forward_extremities.items()
                 for event_id in new_extrem
             ],
@@ -1342,7 +1347,7 @@ class PersistEventsStore:
             d.pop("redacted_because", None)
             return d
 
-        self.db_pool.simple_insert_many_values_txn(
+        self.db_pool.simple_insert_many_txn(
             txn,
             table="event_json",
             keys=("event_id", "room_id", "internal_metadata", "json", "format_version"),
@@ -1358,7 +1363,7 @@ class PersistEventsStore:
             ),
         )
 
-        self.db_pool.simple_insert_many_values_txn(
+        self.db_pool.simple_insert_many_txn(
             txn,
             table="events",
             keys=(
@@ -1412,7 +1417,7 @@ class PersistEventsStore:
         )
         txn.execute(sql + clause, [False] + args)
 
-        self.db_pool.simple_insert_many_values_txn(
+        self.db_pool.simple_insert_many_txn(
             txn,
             table="state_events",
             keys=("event_id", "room_id", "type", "state_key"),
@@ -1622,14 +1627,9 @@ class PersistEventsStore:
         return self.db_pool.simple_insert_many_txn(
             txn=txn,
             table="event_labels",
+            keys=("event_id", "label", "room_id", "topological_ordering"),
             values=[
-                {
-                    "event_id": event_id,
-                    "label": label,
-                    "room_id": room_id,
-                    "topological_ordering": topological_ordering,
-                }
-                for label in labels
+                (event_id, label, room_id, topological_ordering) for label in labels
             ],
         )
 
@@ -1657,16 +1657,13 @@ class PersistEventsStore:
         vals = []
         for event in events:
             ref_alg, ref_hash_bytes = compute_event_reference_hash(event)
-            vals.append(
-                {
-                    "event_id": event.event_id,
-                    "algorithm": ref_alg,
-                    "hash": memoryview(ref_hash_bytes),
-                }
-            )
+            vals.append((event.event_id, ref_alg, memoryview(ref_hash_bytes)))
 
         self.db_pool.simple_insert_many_txn(
-            txn, table="event_reference_hashes", values=vals
+            txn,
+            table="event_reference_hashes",
+            keys=("event_id", "algorithm", "hash"),
+            values=vals,
         )
 
     def _store_room_members_txn(
@@ -1689,18 +1686,25 @@ class PersistEventsStore:
         self.db_pool.simple_insert_many_txn(
             txn,
             table="room_memberships",
+            keys=(
+                "event_id",
+                "user_id",
+                "sender",
+                "room_id",
+                "membership",
+                "display_name",
+                "avatar_url",
+            ),
             values=[
-                {
-                    "event_id": event.event_id,
-                    "user_id": event.state_key,
-                    "sender": event.user_id,
-                    "room_id": event.room_id,
-                    "membership": event.membership,
-                    "display_name": non_null_str_or_none(
-                        event.content.get("displayname")
-                    ),
-                    "avatar_url": non_null_str_or_none(event.content.get("avatar_url")),
-                }
+                (
+                    event.event_id,
+                    event.state_key,
+                    event.user_id,
+                    event.room_id,
+                    event.membership,
+                    non_null_str_or_none(event.content.get("displayname")),
+                    non_null_str_or_none(event.content.get("avatar_url")),
+                )
                 for event in events
             ],
         )
@@ -2163,13 +2167,9 @@ class PersistEventsStore:
         self.db_pool.simple_insert_many_txn(
             txn,
             table="event_edges",
+            keys=("event_id", "prev_event_id", "room_id", "is_state"),
             values=[
-                {
-                    "event_id": ev.event_id,
-                    "prev_event_id": e_id,
-                    "room_id": ev.room_id,
-                    "is_state": False,
-                }
+                (ev.event_id, e_id, ev.room_id, False)
                 for ev in events
                 for e_id in ev.prev_event_ids()
             ],
diff --git a/synapse/storage/databases/main/events_bg_updates.py b/synapse/storage/databases/main/events_bg_updates.py
index 0a96664caf..d5f0059665 100644
--- a/synapse/storage/databases/main/events_bg_updates.py
+++ b/synapse/storage/databases/main/events_bg_updates.py
@@ -684,13 +684,14 @@ class EventsBackgroundUpdatesStore(SQLBaseStore):
                     self.db_pool.simple_insert_many_txn(
                         txn=txn,
                         table="event_labels",
+                        keys=("event_id", "label", "room_id", "topological_ordering"),
                         values=[
-                            {
-                                "event_id": event_id,
-                                "label": label,
-                                "room_id": event_json["room_id"],
-                                "topological_ordering": event_json["depth"],
-                            }
+                            (
+                                event_id,
+                                label,
+                                event_json["room_id"],
+                                event_json["depth"],
+                            )
                             for label in event_json["content"].get(
                                 EventContentFields.LABELS, []
                             )
@@ -803,29 +804,19 @@ class EventsBackgroundUpdatesStore(SQLBaseStore):
 
             if not has_state:
                 state_events.append(
-                    {
-                        "event_id": event.event_id,
-                        "room_id": event.room_id,
-                        "type": event.type,
-                        "state_key": event.state_key,
-                    }
+                    (event.event_id, event.room_id, event.type, event.state_key)
                 )
 
             if not has_event_auth:
                 # Old, dodgy, events may have duplicate auth events, which we
                 # need to deduplicate as we have a unique constraint.
                 for auth_id in set(event.auth_event_ids()):
-                    auth_events.append(
-                        {
-                            "room_id": event.room_id,
-                            "event_id": event.event_id,
-                            "auth_id": auth_id,
-                        }
-                    )
+                    auth_events.append((event.event_id, event.room_id, auth_id))
 
         if state_events:
             await self.db_pool.simple_insert_many(
                 table="state_events",
+                keys=("event_id", "room_id", "type", "state_key"),
                 values=state_events,
                 desc="_rejected_events_metadata_state_events",
             )
@@ -833,6 +824,7 @@ class EventsBackgroundUpdatesStore(SQLBaseStore):
         if auth_events:
             await self.db_pool.simple_insert_many(
                 table="event_auth",
+                keys=("event_id", "room_id", "auth_id"),
                 values=auth_events,
                 desc="_rejected_events_metadata_event_auth",
             )
diff --git a/synapse/storage/databases/main/presence.py b/synapse/storage/databases/main/presence.py
index cbf9ec38f7..4f05811a77 100644
--- a/synapse/storage/databases/main/presence.py
+++ b/synapse/storage/databases/main/presence.py
@@ -129,18 +129,29 @@ class PresenceStore(PresenceBackgroundUpdateStore):
         self.db_pool.simple_insert_many_txn(
             txn,
             table="presence_stream",
+            keys=(
+                "stream_id",
+                "user_id",
+                "state",
+                "last_active_ts",
+                "last_federation_update_ts",
+                "last_user_sync_ts",
+                "status_msg",
+                "currently_active",
+                "instance_name",
+            ),
             values=[
-                {
-                    "stream_id": stream_id,
-                    "user_id": state.user_id,
-                    "state": state.state,
-                    "last_active_ts": state.last_active_ts,
-                    "last_federation_update_ts": state.last_federation_update_ts,
-                    "last_user_sync_ts": state.last_user_sync_ts,
-                    "status_msg": state.status_msg,
-                    "currently_active": state.currently_active,
-                    "instance_name": self._instance_name,
-                }
+                (
+                    stream_id,
+                    state.user_id,
+                    state.state,
+                    state.last_active_ts,
+                    state.last_federation_update_ts,
+                    state.last_user_sync_ts,
+                    state.status_msg,
+                    state.currently_active,
+                    self._instance_name,
+                )
                 for stream_id, state in zip(stream_orderings, presence_states)
             ],
         )
diff --git a/synapse/storage/databases/main/pusher.py b/synapse/storage/databases/main/pusher.py
index 747b4f31df..cf64cd63a4 100644
--- a/synapse/storage/databases/main/pusher.py
+++ b/synapse/storage/databases/main/pusher.py
@@ -561,13 +561,9 @@ class PusherStore(PusherWorkerStore):
             self.db_pool.simple_insert_many_txn(
                 txn,
                 table="deleted_pushers",
+                keys=("stream_id", "app_id", "pushkey", "user_id"),
                 values=[
-                    {
-                        "stream_id": stream_id,
-                        "app_id": pusher.app_id,
-                        "pushkey": pusher.pushkey,
-                        "user_id": user_id,
-                    }
+                    (stream_id, pusher.app_id, pusher.pushkey, user_id)
                     for stream_id, pusher in zip(stream_ids, pushers)
                 ],
             )
diff --git a/synapse/storage/databases/main/user_directory.py b/synapse/storage/databases/main/user_directory.py
index 0f9b8575d3..f7c778bdf2 100644
--- a/synapse/storage/databases/main/user_directory.py
+++ b/synapse/storage/databases/main/user_directory.py
@@ -105,8 +105,10 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore):
                 GROUP BY room_id
             """
             txn.execute(sql)
-            rooms = [{"room_id": x[0], "events": x[1]} for x in txn.fetchall()]
-            self.db_pool.simple_insert_many_txn(txn, TEMP_TABLE + "_rooms", rooms)
+            rooms = list(txn.fetchall())
+            self.db_pool.simple_insert_many_txn(
+                txn, TEMP_TABLE + "_rooms", keys=("room_id", "events"), values=rooms
+            )
             del rooms
 
             sql = (
@@ -117,9 +119,11 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore):
             txn.execute(sql)
 
             txn.execute("SELECT name FROM users")
-            users = [{"user_id": x[0]} for x in txn.fetchall()]
+            users = list(txn.fetchall())
 
-            self.db_pool.simple_insert_many_txn(txn, TEMP_TABLE + "_users", users)
+            self.db_pool.simple_insert_many_txn(
+                txn, TEMP_TABLE + "_users", keys=("user_id",), values=users
+            )
 
         new_pos = await self.get_max_stream_id_in_current_state_deltas()
         await self.db_pool.runInteraction(
diff --git a/synapse/storage/databases/state/bg_updates.py b/synapse/storage/databases/state/bg_updates.py
index eb1118d2cb..5de70f31d2 100644
--- a/synapse/storage/databases/state/bg_updates.py
+++ b/synapse/storage/databases/state/bg_updates.py
@@ -327,14 +327,15 @@ class StateBackgroundUpdateStore(StateGroupBackgroundUpdateStore):
                         self.db_pool.simple_insert_many_txn(
                             txn,
                             table="state_groups_state",
+                            keys=(
+                                "state_group",
+                                "room_id",
+                                "type",
+                                "state_key",
+                                "event_id",
+                            ),
                             values=[
-                                {
-                                    "state_group": state_group,
-                                    "room_id": room_id,
-                                    "type": key[0],
-                                    "state_key": key[1],
-                                    "event_id": state_id,
-                                }
+                                (state_group, room_id, key[0], key[1], state_id)
                                 for key, state_id in delta_state.items()
                             ],
                         )
diff --git a/synapse/storage/databases/state/store.py b/synapse/storage/databases/state/store.py
index c4c8c0021b..7614d76ac6 100644
--- a/synapse/storage/databases/state/store.py
+++ b/synapse/storage/databases/state/store.py
@@ -460,14 +460,9 @@ class StateGroupDataStore(StateBackgroundUpdateStore, SQLBaseStore):
                 self.db_pool.simple_insert_many_txn(
                     txn,
                     table="state_groups_state",
+                    keys=("state_group", "room_id", "type", "state_key", "event_id"),
                     values=[
-                        {
-                            "state_group": state_group,
-                            "room_id": room_id,
-                            "type": key[0],
-                            "state_key": key[1],
-                            "event_id": state_id,
-                        }
+                        (state_group, room_id, key[0], key[1], state_id)
                         for key, state_id in delta_ids.items()
                     ],
                 )
@@ -475,14 +470,9 @@ class StateGroupDataStore(StateBackgroundUpdateStore, SQLBaseStore):
                 self.db_pool.simple_insert_many_txn(
                     txn,
                     table="state_groups_state",
+                    keys=("state_group", "room_id", "type", "state_key", "event_id"),
                     values=[
-                        {
-                            "state_group": state_group,
-                            "room_id": room_id,
-                            "type": key[0],
-                            "state_key": key[1],
-                            "event_id": state_id,
-                        }
+                        (state_group, room_id, key[0], key[1], state_id)
                         for key, state_id in current_state_ids.items()
                     ],
                 )
@@ -589,14 +579,9 @@ class StateGroupDataStore(StateBackgroundUpdateStore, SQLBaseStore):
             self.db_pool.simple_insert_many_txn(
                 txn,
                 table="state_groups_state",
+                keys=("state_group", "room_id", "type", "state_key", "event_id"),
                 values=[
-                    {
-                        "state_group": sg,
-                        "room_id": room_id,
-                        "type": key[0],
-                        "state_key": key[1],
-                        "event_id": state_id,
-                    }
+                    (sg, room_id, key[0], key[1], state_id)
                     for key, state_id in curr_state.items()
                 ],
             )
diff --git a/tests/rest/admin/test_registration_tokens.py b/tests/rest/admin/test_registration_tokens.py
index 81f3ac7f04..8513b1d2df 100644
--- a/tests/rest/admin/test_registration_tokens.py
+++ b/tests/rest/admin/test_registration_tokens.py
@@ -223,20 +223,13 @@ class ManageRegistrationTokensTestCase(unittest.HomeserverTestCase):
         # Create all possible single character tokens
         tokens = []
         for c in string.ascii_letters + string.digits + "._~-":
-            tokens.append(
-                {
-                    "token": c,
-                    "uses_allowed": None,
-                    "pending": 0,
-                    "completed": 0,
-                    "expiry_time": None,
-                }
-            )
+            tokens.append((c, None, 0, 0, None))
         self.get_success(
             self.store.db_pool.simple_insert_many(
                 "registration_tokens",
-                tokens,
-                "create_all_registration_tokens",
+                keys=("token", "uses_allowed", "pending", "completed", "expiry_time"),
+                values=tokens,
+                desc="create_all_registration_tokens",
             )
         )
 
diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py
index ecfda7677e..632bbc9de7 100644
--- a/tests/storage/test_event_federation.py
+++ b/tests/storage/test_event_federation.py
@@ -515,17 +515,23 @@ class EventFederationWorkerStoreTestCase(tests.unittest.HomeserverTestCase):
         self.get_success(
             self.store.db_pool.simple_insert_many(
                 table="federation_inbound_events_staging",
+                keys=(
+                    "origin",
+                    "room_id",
+                    "received_ts",
+                    "event_id",
+                    "event_json",
+                    "internal_metadata",
+                ),
                 values=[
-                    {
-                        "origin": "some_origin",
-                        "room_id": room_id,
-                        "received_ts": 0,
-                        "event_id": f"$fake_event_id_{i + 1}",
-                        "event_json": json_encoder.encode(
-                            {"prev_events": [f"$fake_event_id_{i}"]}
-                        ),
-                        "internal_metadata": "{}",
-                    }
+                    (
+                        "some_origin",
+                        room_id,
+                        0,
+                        f"$fake_event_id_{i + 1}",
+                        json_encoder.encode({"prev_events": [f"$fake_event_id_{i}"]}),
+                        "{}",
+                    )
                     for i in range(500)
                 ],
                 desc="test_prune_inbound_federation_queue",