diff --git a/swh/provenance/archive.py b/swh/provenance/archive.py
--- a/swh/provenance/archive.py
+++ b/swh/provenance/archive.py
@@ -2,12 +2,12 @@
 
 from typing_extensions import Protocol, runtime_checkable
 
-from swh.model.model import Revision, Sha1
+from swh.model.model import Revision, Sha1Git
 
 
 @runtime_checkable
 class ArchiveInterface(Protocol):
-    def directory_ls(self, id: Sha1) -> Iterable[Dict[str, Any]]:
+    def directory_ls(self, id: Sha1Git) -> Iterable[Dict[str, Any]]:
         """List entries for one directory.
 
         Args:
@@ -19,7 +19,7 @@
         """
         ...
 
-    def revision_get(self, ids: Iterable[Sha1]) -> Iterable[Revision]:
+    def revision_get(self, ids: Iterable[Sha1Git]) -> Iterable[Revision]:
         """Given a list of sha1, return the revisions' information
 
         Args:
@@ -32,11 +32,11 @@
         """
         ...
 
-    def snapshot_get_heads(self, id: Sha1) -> Iterable[Sha1]:
+    def snapshot_get_heads(self, id: Sha1Git) -> Iterable[Sha1Git]:
         """List all revisions pointed by one snapshot.
 
         Args:
-            snapshot: the snapshot's identifier
+            id: sha1 id of the snapshot.
 
         Yields:
             sha1 ids of found revisions.
diff --git a/swh/provenance/graph.py b/swh/provenance/graph.py
--- a/swh/provenance/graph.py
+++ b/swh/provenance/graph.py
@@ -3,6 +3,8 @@
 import os
 from typing import Dict, Optional, Set
 
+from swh.model.model import Sha1Git
+
 from .archive import ArchiveInterface
 from .model import DirectoryEntry, RevisionEntry
 from .provenance import ProvenanceInterface
@@ -166,7 +168,7 @@
     logging.debug(
         f"Recursively creating isochrone graph for revision {revision.id.hex()}..."
     )
-    fdates: Dict[bytes, datetime] = {}  # map {file_id: date}
+    fdates: Dict[Sha1Git, datetime] = {}  # map {file_id: date}
     while stack:
         current = stack.pop()
         if current.dbdate is None or current.dbdate > revision.date:
diff --git a/swh/provenance/model.py b/swh/provenance/model.py
--- a/swh/provenance/model.py
+++ b/swh/provenance/model.py
@@ -6,14 +6,17 @@
 from datetime import datetime
 from typing import Iterable, Iterator, List, Optional
 
+from swh.model.hashutil import hash_to_bytes
+from swh.model.identifiers import origin_identifier
+from swh.model.model import Sha1Git
+
 from .archive import ArchiveInterface
 
 
 class OriginEntry:
-    def __init__(self, url: str, date: datetime, snapshot: bytes):
+    def __init__(self, url: str, snapshot: Sha1Git):
         self.url = url
-        # TODO: this is probably not needed and will be removed!
-        # self.date = date
+        self.id: Sha1Git = hash_to_bytes(origin_identifier({"url": self.url}))
         self.snapshot = snapshot
         self._revisions: Optional[List[RevisionEntry]] = None
 
@@ -33,19 +36,16 @@
         return (x for x in self._revisions)
 
     def __str__(self):
-        return (
-            f"<MOrigin[{self.url}] "
-            f"snap={self.snapshot.hex()}, date={self.date.isoformat()}>"
-        )
+        return f"<MOrigin[{self.id.hex()}] url={self.url}, snap={self.snapshot.hex()}>"
 
 
 class RevisionEntry:
     def __init__(
         self,
-        id: bytes,
+        id: Sha1Git,
         date: Optional[datetime] = None,
-        root: Optional[bytes] = None,
-        parents: Optional[Iterable[bytes]] = None,
+        root: Optional[Sha1Git] = None,
+        parents: Optional[Iterable[Sha1Git]] = None,
     ):
         self.id = id
         self.date = date
@@ -91,7 +91,7 @@
 
 
 class DirectoryEntry:
-    def __init__(self, id: bytes, name: bytes = b""):
+    def __init__(self, id: Sha1Git, name: bytes = b""):
         self.id = id
         self.name = name
         self._files: Optional[List[FileEntry]] = None
@@ -141,7 +141,7 @@
 
 
 class FileEntry:
-    def __init__(self, id: bytes, name: bytes):
+    def __init__(self, id: Sha1Git, name: bytes):
         self.id = id
         self.name = name
 
diff --git a/swh/provenance/origin.py b/swh/provenance/origin.py
--- a/swh/provenance/origin.py
+++ b/swh/provenance/origin.py
@@ -1,10 +1,9 @@
-from datetime import datetime, timezone
 from itertools import islice
 import logging
 import time
 from typing import Iterable, Iterator, List, Optional, Tuple
 
-import iso8601
+from swh.model.model import Sha1Git
 
 from .archive import ArchiveInterface
 from .graph import HistoryNode, build_history_graph
@@ -16,31 +15,28 @@
     """Iterator over origin visit statuses typically present in the given CSV
     file.
 
-    The input is an iterator that produces 3 elements per row:
+    The input is an iterator that produces 2 elements per row:
 
-      (url, date, snap)
+      (url, snap)
 
     where:
     - url: is the origin url of the visit
-    - date: is the date of the visit
     - snap: sha1_git of the snapshot pointed by the visit status
     """
 
     def __init__(
         self,
-        statuses: Iterable[Tuple[str, datetime, bytes]],
+        statuses: Iterable[Tuple[str, Sha1Git]],
         limit: Optional[int] = None,
     ):
-        self.statuses: Iterator[Tuple[str, datetime, bytes]]
+        self.statuses: Iterator[Tuple[str, Sha1Git]]
         if limit is not None:
             self.statuses = islice(statuses, limit)
         else:
             self.statuses = iter(statuses)
 
     def __iter__(self):
-        for url, date, snap in self.statuses:
-            date = iso8601.parse_date(date, default_timezone=timezone.utc)
-            yield OriginEntry(url, date, snap)
+        return (OriginEntry(url, snapshot) for url, snapshot in self.statuses)
 
 
 def origin_add(
@@ -50,6 +46,7 @@
 ):
     start = time.time()
     for origin in origins:
+        provenance.origin_add(origin)
         origin.retrieve_revisions(archive)
         for revision in origin.revisions:
             graph = build_history_graph(archive, provenance, revision)
@@ -59,7 +56,7 @@
     stop = time.time()
     logging.debug(
         "Origins "
-        ";".join([origin.url + ":" + origin.snapshot.hex() for origin in origins])
+        ";".join([origin.id.hex() + ":" + origin.snapshot.hex() for origin in origins])
         + f" were processed in {stop - start} secs (commit took {stop - done} secs)!"
     )
 
diff --git a/swh/provenance/postgresql/archive.py b/swh/provenance/postgresql/archive.py
--- a/swh/provenance/postgresql/archive.py
+++ b/swh/provenance/postgresql/archive.py
@@ -3,7 +3,7 @@
 from methodtools import lru_cache
 import psycopg2
 
-from swh.model.model import ObjectType, Revision, Sha1, TargetType
+from swh.model.model import ObjectType, Revision, Sha1Git, TargetType
 from swh.storage.postgresql.storage import Storage
 
 
@@ -12,14 +12,14 @@
         self.conn = conn
         self.storage = Storage(conn, objstorage={"cls": "memory"})
 
-    def directory_ls(self, id: Sha1) -> Iterable[Dict[str, Any]]:
+    def directory_ls(self, id: Sha1Git) -> Iterable[Dict[str, Any]]:
         # TODO: only call directory_ls_internal if the id is not being queried by
         # someone else. Otherwise wait until results get properly cached.
         entries = self.directory_ls_internal(id)
         yield from entries
 
     @lru_cache(maxsize=100000)
-    def directory_ls_internal(self, id: Sha1) -> List[Dict[str, Any]]:
+    def directory_ls_internal(self, id: Sha1Git) -> List[Dict[str, Any]]:
         # TODO: add file size filtering
         with self.conn.cursor() as cursor:
             cursor.execute(
@@ -62,7 +62,7 @@
                 for row in cursor.fetchall()
             ]
 
-    def revision_get(self, ids: Iterable[Sha1]) -> Iterable[Revision]:
+    def revision_get(self, ids: Iterable[Sha1Git]) -> Iterable[Revision]:
         with self.conn.cursor() as cursor:
             psycopg2.extras.execute_values(
                 cursor,
@@ -96,7 +96,7 @@
                     }
                 )
 
-    def snapshot_get_heads(self, id: Sha1) -> Iterable[Sha1]:
+    def snapshot_get_heads(self, id: Sha1Git) -> Iterable[Sha1Git]:
         # TODO: this code is duplicated here (same as in swh.provenance.storage.archive)
         # but it's just temporary. This method should actually perform a direct query to
         # the SQL db of the archive.
@@ -123,7 +123,7 @@
                 if release is not None and release.target_type == ObjectType.REVISION
             )
 
-        revisions: Set[Sha1] = set()
+        revisions: Set[Sha1Git] = set()
         for targets in grouper(targets_set, batchsize):
             revisions.update(
                 revision.id
diff --git a/swh/provenance/postgresql/provenancedb_base.py b/swh/provenance/postgresql/provenancedb_base.py
--- a/swh/provenance/postgresql/provenancedb_base.py
+++ b/swh/provenance/postgresql/provenancedb_base.py
@@ -6,6 +6,8 @@
 import psycopg2
 import psycopg2.extras
 
+from swh.model.model import Sha1Git
+
 
 class ProvenanceDBBase:
     def __init__(self, conn: psycopg2.extensions.connection):
@@ -51,6 +53,16 @@
             ):
                 self.insert_relation(relation, data[relation])
 
+            # Insert origins
+            self.insert_origin(
+                {
+                    sha1: data["origin"]["data"][sha1]
+                    for sha1 in data["origin"]["added"]
+                },
+            )
+            data["origin"]["data"].clear()
+            data["origin"]["added"].clear()
+
             # Insert relations from the origin-revision layer
             self.insert_origin_head(data["revision_in_origin"])
             self.insert_revision_history(data["revision_before_revision"])
@@ -58,12 +70,12 @@
             # Update preferred origins
             self.update_preferred_origin(
                 {
-                    sha1: data["revision_preferred_origin"]["data"][sha1]
-                    for sha1 in data["revision_preferred_origin"]["added"]
+                    sha1: data["revision_origin"]["data"][sha1]
+                    for sha1 in data["revision_origin"]["added"]
                 }
             )
-            data["revision_preferred_origin"]["data"].clear()
-            data["revision_preferred_origin"]["added"].clear()
+            data["revision_origin"]["data"].clear()
+            data["revision_origin"]["added"].clear()
 
             return True
 
@@ -76,16 +88,16 @@
         return False
 
     def content_find_first(
-        self, blob: bytes
-    ) -> Optional[Tuple[bytes, bytes, datetime, bytes]]:
+        self, id: Sha1Git
+    ) -> Optional[Tuple[Sha1Git, Sha1Git, datetime, bytes]]:
         ...
 
     def content_find_all(
-        self, blob: bytes, limit: Optional[int] = None
-    ) -> Generator[Tuple[bytes, bytes, datetime, bytes], None, None]:
+        self, id: Sha1Git, limit: Optional[int] = None
+    ) -> Generator[Tuple[Sha1Git, Sha1Git, datetime, bytes], None, None]:
         ...
 
-    def get_dates(self, entity: str, ids: List[bytes]) -> Dict[bytes, datetime]:
+    def get_dates(self, entity: str, ids: List[Sha1Git]) -> Dict[Sha1Git, datetime]:
         dates = {}
         if ids:
             values = ", ".join(itertools.repeat("%s", len(ids)))
@@ -96,7 +108,7 @@
             dates.update(self.cursor.fetchall())
         return dates
 
-    def insert_entity(self, entity: str, data: Dict[bytes, datetime]):
+    def insert_entity(self, entity: str, data: Dict[Sha1Git, datetime]):
         if data:
             psycopg2.extras.execute_values(
                 self.cursor,
@@ -112,7 +124,22 @@
             #      This might be useless!
             data.clear()
 
-    def insert_origin_head(self, data: Set[Tuple[bytes, str]]):
+    def insert_origin(self, data: Dict[Sha1Git, str]):
+        if data:
+            psycopg2.extras.execute_values(
+                self.cursor,
+                """
+                LOCK TABLE ONLY origin;
+                INSERT INTO origin(sha1, url) VALUES %s
+                  ON CONFLICT DO NOTHING
+                """,
+                data.items(),
+            )
+            # XXX: not sure if Python takes a reference or a copy.
+            #      This might be useless!
+            data.clear()
+
+    def insert_origin_head(self, data: Set[Tuple[Sha1Git, Sha1Git]]):
         if data:
             psycopg2.extras.execute_values(
                 self.cursor,
@@ -123,16 +150,16 @@
                     SELECT R.id, O.id
                     FROM (VALUES %s) AS V(rev, org)
                     INNER JOIN revision AS R on (R.sha1=V.rev)
-                    INNER JOIN origin AS O on (O.url=V.org::unix_path)
+                    INNER JOIN origin AS O on (O.sha1=V.org)
                 """,
                 data,
             )
             data.clear()
 
-    def insert_relation(self, relation: str, data: Set[Tuple[bytes, bytes, bytes]]):
+    def insert_relation(self, relation: str, data: Set[Tuple[Sha1Git, Sha1Git, bytes]]):
         ...
 
-    def insert_revision_history(self, data: Dict[bytes, bytes]):
+    def insert_revision_history(self, data: Dict[Sha1Git, Sha1Git]):
         if data:
             values = [[(prev, next) for next in data[prev]] for prev in data]
             psycopg2.extras.execute_values(
@@ -150,10 +177,10 @@
             )
             data.clear()
 
-    def revision_get_preferred_origin(self, revision: bytes) -> Optional[str]:
+    def revision_get_preferred_origin(self, revision: Sha1Git) -> Optional[Sha1Git]:
         self.cursor.execute(
             """
-            SELECT O.url
+            SELECT O.sha1
               FROM revision AS R
               JOIN origin as O
                 ON R.origin=O.id
@@ -161,9 +188,9 @@
             (revision,),
         )
         row = self.cursor.fetchone()
-        return str(row[0], encoding="utf-8") if row is not None else None
+        return row[0] if row is not None else None
 
-    def revision_in_history(self, revision: bytes) -> bool:
+    def revision_in_history(self, revision: Sha1Git) -> bool:
         self.cursor.execute(
             """
             SELECT 1
@@ -176,7 +203,7 @@
         )
         return self.cursor.fetchone() is not None
 
-    def revision_visited(self, revision: bytes) -> bool:
+    def revision_visited(self, revision: Sha1Git) -> bool:
         self.cursor.execute(
             """
             SELECT 1
@@ -189,18 +216,18 @@
         )
         return self.cursor.fetchone() is not None
 
-    def update_preferred_origin(self, data: Dict[bytes, str]):
+    def update_preferred_origin(self, data: Dict[Sha1Git, Sha1Git]):
         if data:
             # XXX: this is assuming the revision already exists in the db! It should
             #      be improved by allowing null dates in the revision table.
             psycopg2.extras.execute_values(
                 self.cursor,
                 """
-                UPDATE revision
+                UPDATE revision R
                   SET origin=O.id
                   FROM (VALUES %s) AS V(rev, org)
-                  INNER JOIN origin AS O on (O.url=V.org::unix_path)
-                  WHERE sha1=V.rev
+                  INNER JOIN origin AS O on (O.sha1=V.org)
+                  WHERE R.sha1=V.rev
                 """,
                 data.items(),
             )
diff --git a/swh/provenance/postgresql/provenancedb_with_path.py b/swh/provenance/postgresql/provenancedb_with_path.py
--- a/swh/provenance/postgresql/provenancedb_with_path.py
+++ b/swh/provenance/postgresql/provenancedb_with_path.py
@@ -4,13 +4,15 @@
 import psycopg2
 import psycopg2.extras
 
+from swh.model.model import Sha1Git
+
 from .provenancedb_base import ProvenanceDBBase
 
 
 class ProvenanceWithPathDB(ProvenanceDBBase):
     def content_find_first(
-        self, blob: bytes
-    ) -> Optional[Tuple[bytes, bytes, datetime, bytes]]:
+        self, id: Sha1Git
+    ) -> Optional[Tuple[Sha1Git, Sha1Git, datetime, bytes]]:
         self.cursor.execute(
             """
             SELECT C.sha1 AS blob,
@@ -24,13 +26,13 @@
             WHERE C.sha1=%s
             ORDER BY date, rev, path ASC LIMIT 1
             """,
-            (blob,),
+            (id,),
         )
         return self.cursor.fetchone()
 
     def content_find_all(
-        self, blob: bytes, limit: Optional[int] = None
-    ) -> Generator[Tuple[bytes, bytes, datetime, bytes], None, None]:
+        self, id: Sha1Git, limit: Optional[int] = None
+    ) -> Generator[Tuple[Sha1Git, Sha1Git, datetime, bytes], None, None]:
         early_cut = f"LIMIT {limit}" if limit is not None else ""
         self.cursor.execute(
             f"""
@@ -61,11 +63,11 @@
              WHERE C.sha1=%s)
             ORDER BY date, rev, path {early_cut}
             """,
-            (blob, blob),
+            (id, id),
         )
         yield from self.cursor.fetchall()
 
-    def insert_relation(self, relation: str, data: Set[Tuple[bytes, bytes, bytes]]):
+    def insert_relation(self, relation: str, data: Set[Tuple[Sha1Git, Sha1Git, bytes]]):
         """Insert entries in `relation` from `data`
 
         Also insert missing location entries in the 'location' table.
diff --git a/swh/provenance/postgresql/provenancedb_without_path.py b/swh/provenance/postgresql/provenancedb_without_path.py
--- a/swh/provenance/postgresql/provenancedb_without_path.py
+++ b/swh/provenance/postgresql/provenancedb_without_path.py
@@ -4,17 +4,15 @@
 import psycopg2
 import psycopg2.extras
 
-from .provenancedb_base import ProvenanceDBBase
+from swh.model.model import Sha1Git
 
-########################################################################################
-########################################################################################
-########################################################################################
+from .provenancedb_base import ProvenanceDBBase
 
 
 class ProvenanceWithoutPathDB(ProvenanceDBBase):
     def content_find_first(
-        self, blob: bytes
-    ) -> Optional[Tuple[bytes, bytes, datetime, bytes]]:
+        self, id: Sha1Git
+    ) -> Optional[Tuple[Sha1Git, Sha1Git, datetime, bytes]]:
         self.cursor.execute(
             """
             SELECT C.sha1 AS blob,
@@ -27,13 +25,13 @@
             WHERE C.sha1=%s
             ORDER BY date, rev ASC LIMIT 1
             """,
-            (blob,),
+            (id,),
         )
         return self.cursor.fetchone()
 
     def content_find_all(
-        self, blob: bytes, limit: Optional[int] = None
-    ) -> Generator[Tuple[bytes, bytes, datetime, bytes], None, None]:
+        self, id: Sha1Git, limit: Optional[int] = None
+    ) -> Generator[Tuple[Sha1Git, Sha1Git, datetime, bytes], None, None]:
         early_cut = f"LIMIT {limit}" if limit is not None else ""
         self.cursor.execute(
             f"""
@@ -57,11 +55,11 @@
              WHERE C.sha1=%s)
             ORDER BY date, rev, path {early_cut}
             """,
-            (blob, blob),
+            (id, id),
         )
         yield from self.cursor.fetchall()
 
-    def insert_relation(self, relation: str, data: Set[Tuple[bytes, bytes, bytes]]):
+    def insert_relation(self, relation: str, data: Set[Tuple[Sha1Git, Sha1Git, bytes]]):
         if data:
             assert relation in (
                 "content_in_revision",
diff --git a/swh/provenance/provenance.py b/swh/provenance/provenance.py
--- a/swh/provenance/provenance.py
+++ b/swh/provenance/provenance.py
@@ -6,6 +6,8 @@
 import psycopg2
 from typing_extensions import Literal, Protocol, TypedDict, runtime_checkable
 
+from swh.model.model import Sha1Git
+
 from .model import DirectoryEntry, FileEntry, OriginEntry, RevisionEntry
 
 
@@ -30,13 +32,13 @@
         ...
 
     def content_find_first(
-        self, blob: bytes
-    ) -> Optional[Tuple[bytes, bytes, datetime, bytes]]:
+        self, id: Sha1Git
+    ) -> Optional[Tuple[Sha1Git, Sha1Git, datetime, bytes]]:
         ...
 
     def content_find_all(
-        self, blob: bytes, limit: Optional[int] = None
-    ) -> Generator[Tuple[bytes, bytes, datetime, bytes], None, None]:
+        self, id: Sha1Git, limit: Optional[int] = None
+    ) -> Generator[Tuple[Sha1Git, Sha1Git, datetime, bytes], None, None]:
         ...
 
     def content_get_early_date(self, blob: FileEntry) -> Optional[datetime]:
@@ -44,7 +46,7 @@
 
     def content_get_early_dates(
         self, blobs: Iterable[FileEntry]
-    ) -> Dict[bytes, datetime]:
+    ) -> Dict[Sha1Git, datetime]:
         ...
 
     def content_set_early_date(self, blob: FileEntry, date: datetime) -> None:
@@ -62,7 +64,7 @@
 
     def directory_get_dates_in_isochrone_frontier(
         self, dirs: Iterable[DirectoryEntry]
-    ) -> Dict[bytes, datetime]:
+    ) -> Dict[Sha1Git, datetime]:
         ...
 
     def directory_set_date_in_isochrone_frontier(
@@ -70,6 +72,9 @@
     ) -> None:
         ...
 
+    def origin_add(self, origin: OriginEntry) -> None:
+        ...
+
     def revision_add(self, revision: RevisionEntry) -> None:
         ...
 
@@ -86,7 +91,9 @@
     def revision_get_early_date(self, revision: RevisionEntry) -> Optional[datetime]:
         ...
 
-    def revision_get_preferred_origin(self, revision: RevisionEntry) -> Optional[str]:
+    def revision_get_preferred_origin(
+        self, revision: RevisionEntry
+    ) -> Optional[Sha1Git]:
         ...
 
     def revision_in_history(self, revision: RevisionEntry) -> bool:
@@ -102,13 +109,18 @@
 
 
 class DatetimeCache(TypedDict):
-    data: Dict[bytes, datetime]
-    added: Set[bytes]
+    data: Dict[Sha1Git, datetime]
+    added: Set[Sha1Git]
 
 
 class OriginCache(TypedDict):
-    data: Dict[bytes, str]
-    added: Set[bytes]
+    data: Dict[Sha1Git, str]
+    added: Set[Sha1Git]
+
+
+class RevisionCache(TypedDict):
+    data: Dict[Sha1Git, Sha1Git]
+    added: Set[Sha1Git]
 
 
 class ProvenanceCache(TypedDict):
@@ -116,13 +128,14 @@
     directory: DatetimeCache
     revision: DatetimeCache
     # below are insertion caches only
-    content_in_revision: Set[Tuple[bytes, bytes, bytes]]
-    content_in_directory: Set[Tuple[bytes, bytes, bytes]]
-    directory_in_revision: Set[Tuple[bytes, bytes, bytes]]
+    content_in_revision: Set[Tuple[Sha1Git, Sha1Git, bytes]]
+    content_in_directory: Set[Tuple[Sha1Git, Sha1Git, bytes]]
+    directory_in_revision: Set[Tuple[Sha1Git, Sha1Git, bytes]]
     # these two are for the origin layer
-    revision_before_revision: Dict[bytes, Set[bytes]]
-    revision_in_origin: Set[Tuple[bytes, str]]
-    revision_preferred_origin: OriginCache
+    origin: OriginCache
+    revision_origin: RevisionCache
+    revision_before_revision: Dict[Sha1Git, Set[Sha1Git]]
+    revision_in_origin: Set[Tuple[Sha1Git, Sha1Git]]
 
 
 def new_cache():
@@ -133,9 +146,10 @@
         content_in_revision=set(),
         content_in_directory=set(),
         directory_in_revision=set(),
+        origin=OriginCache(data={}, added=set()),
+        revision_origin=RevisionCache(data={}, added=set()),
         revision_before_revision={},
         revision_in_origin=set(),
-        revision_preferred_origin=OriginCache(data={}, added=set()),
     )
 
 
@@ -185,21 +199,21 @@
         )
 
     def content_find_first(
-        self, blob: bytes
-    ) -> Optional[Tuple[bytes, bytes, datetime, bytes]]:
-        return self.storage.content_find_first(blob)
+        self, id: Sha1Git
+    ) -> Optional[Tuple[Sha1Git, Sha1Git, datetime, bytes]]:
+        return self.storage.content_find_first(id)
 
     def content_find_all(
-        self, blob: bytes, limit: Optional[int] = None
-    ) -> Generator[Tuple[bytes, bytes, datetime, bytes], None, None]:
-        yield from self.storage.content_find_all(blob, limit=limit)
+        self, id: Sha1Git, limit: Optional[int] = None
+    ) -> Generator[Tuple[Sha1Git, Sha1Git, datetime, bytes], None, None]:
+        yield from self.storage.content_find_all(id, limit=limit)
 
     def content_get_early_date(self, blob: FileEntry) -> Optional[datetime]:
         return self.get_dates("content", [blob.id]).get(blob.id, None)
 
     def content_get_early_dates(
         self, blobs: Iterable[FileEntry]
-    ) -> Dict[bytes, datetime]:
+    ) -> Dict[Sha1Git, datetime]:
         return self.get_dates("content", [blob.id for blob in blobs])
 
     def content_set_early_date(self, blob: FileEntry, date: datetime):
@@ -220,7 +234,7 @@
 
     def directory_get_dates_in_isochrone_frontier(
         self, dirs: Iterable[DirectoryEntry]
-    ) -> Dict[bytes, datetime]:
+    ) -> Dict[Sha1Git, datetime]:
         return self.get_dates("directory", [directory.id for directory in dirs])
 
     def directory_set_date_in_isochrone_frontier(
@@ -230,14 +244,18 @@
         self.cache["directory"]["added"].add(directory.id)
 
     def get_dates(
-        self, entity: Literal["content", "revision", "directory"], ids: List[bytes]
-    ) -> Dict[bytes, datetime]:
+        self, entity: Literal["content", "revision", "directory"], ids: List[Sha1Git]
+    ) -> Dict[Sha1Git, datetime]:
         cache = self.cache[entity]
         missing_ids = set(id for id in ids if id not in cache)
         if missing_ids:
             cache["data"].update(self.storage.get_dates(entity, list(missing_ids)))
         return {sha1: cache["data"][sha1] for sha1 in ids if sha1 in cache["data"]}
 
+    def origin_add(self, origin: OriginEntry) -> None:
+        self.cache["origin"]["data"][origin.id] = origin.url
+        self.cache["origin"]["added"].add(origin.id)
+
     def revision_add(self, revision: RevisionEntry):
         # Add current revision to the compact DB
         assert revision.date is not None
@@ -252,17 +270,20 @@
         )
 
     def revision_add_to_origin(self, origin: OriginEntry, revision: RevisionEntry):
-        self.cache["revision_in_origin"].add((revision.id, origin.url))
+        self.cache["revision_in_origin"].add((revision.id, origin.id))
 
     def revision_get_early_date(self, revision: RevisionEntry) -> Optional[datetime]:
         return self.get_dates("revision", [revision.id]).get(revision.id, None)
 
-    def revision_get_preferred_origin(self, revision: RevisionEntry) -> Optional[str]:
-        if revision.id not in self.cache["revision_preferred_origin"]["data"]:
-            url = self.storage.revision_get_preferred_origin(revision.id)
-            if url is not None:
-                self.cache["revision_preferred_origin"]["data"][revision.id] = url
-        return self.cache["revision_preferred_origin"]["data"].get(revision.id)
+    def revision_get_preferred_origin(
+        self, revision: RevisionEntry
+    ) -> Optional[Sha1Git]:
+        cache = self.cache["revision_origin"]
+        if revision.id not in cache:
+            origin = self.storage.revision_get_preferred_origin(revision.id)
+            if origin is not None:
+                cache["data"][revision.id] = origin
+        return cache["data"].get(revision.id)
 
     def revision_in_history(self, revision: RevisionEntry) -> bool:
         return revision.id in self.cache[
@@ -272,8 +293,8 @@
     def revision_set_preferred_origin(
         self, origin: OriginEntry, revision: RevisionEntry
     ):
-        self.cache["revision_preferred_origin"]["data"][revision.id] = origin.url
-        self.cache["revision_preferred_origin"]["added"].add(revision.id)
+        self.cache["revision_origin"]["data"][revision.id] = origin.id
+        self.cache["revision_origin"]["added"].add(revision.id)
 
     def revision_visited(self, revision: RevisionEntry) -> bool:
         return revision.id in dict(
diff --git a/swh/provenance/revision.py b/swh/provenance/revision.py
--- a/swh/provenance/revision.py
+++ b/swh/provenance/revision.py
@@ -8,6 +8,7 @@
 import iso8601
 
 from swh.model.hashutil import hash_to_bytes
+from swh.model.model import Sha1Git
 
 from .archive import ArchiveInterface
 from .graph import IsochroneNode, build_isochrone_graph
@@ -30,10 +31,10 @@
 
     def __init__(
         self,
-        revisions: Iterable[Tuple[bytes, datetime, bytes]],
+        revisions: Iterable[Tuple[Sha1Git, datetime, Sha1Git]],
         limit: Optional[int] = None,
     ):
-        self.revisions: Iterator[Tuple[bytes, datetime, bytes]]
+        self.revisions: Iterator[Tuple[Sha1Git, datetime, Sha1Git]]
         if limit is not None:
             self.revisions = islice(revisions, limit)
         else:
diff --git a/swh/provenance/sql/30-schema.sql b/swh/provenance/sql/30-schema.sql
--- a/swh/provenance/sql/30-schema.sql
+++ b/swh/provenance/sql/30-schema.sql
@@ -67,9 +67,11 @@
 create table origin
 (
     id      bigserial primary key,      -- internal identifier of the origin
+    sha1    sha1_git unique not null,   -- intrinsic identifier of the origin
     url     unix_path unique not null   -- url of the origin
 );
 comment on column origin.id is 'Origin internal identifier';
+comment on column origin.sha1 is 'Origin intrinsic identifier';
 comment on column origin.url is 'URL of the origin';
 
 -- relation tables
diff --git a/swh/provenance/storage/archive.py b/swh/provenance/storage/archive.py
--- a/swh/provenance/storage/archive.py
+++ b/swh/provenance/storage/archive.py
@@ -1,6 +1,6 @@
 from typing import Any, Dict, Iterable, Set
 
-from swh.model.model import ObjectType, Revision, Sha1, TargetType
+from swh.model.model import ObjectType, Revision, Sha1Git, TargetType
 from swh.storage.interface import StorageInterface
 
 
@@ -8,17 +8,17 @@
     def __init__(self, storage: StorageInterface):
         self.storage = storage
 
-    def directory_ls(self, id: Sha1) -> Iterable[Dict[str, Any]]:
+    def directory_ls(self, id: Sha1Git) -> Iterable[Dict[str, Any]]:
         # TODO: filter unused fields
         yield from self.storage.directory_ls(id)
 
-    def revision_get(self, ids: Iterable[Sha1]) -> Iterable[Revision]:
+    def revision_get(self, ids: Iterable[Sha1Git]) -> Iterable[Revision]:
         # TODO: filter unused fields
         yield from (
             rev for rev in self.storage.revision_get(list(ids)) if rev is not None
         )
 
-    def snapshot_get_heads(self, id: Sha1) -> Iterable[Sha1]:
+    def snapshot_get_heads(self, id: Sha1Git) -> Iterable[Sha1Git]:
         from swh.core.utils import grouper
         from swh.storage.algos.snapshot import snapshot_get_all_branches
 
@@ -42,7 +42,7 @@
                 if release is not None and release.target_type == ObjectType.REVISION
             )
 
-        revisions: Set[Sha1] = set()
+        revisions: Set[Sha1Git] = set()
         for targets in grouper(targets_set, batchsize):
             revisions.update(
                 revision.id
diff --git a/swh/provenance/tests/conftest.py b/swh/provenance/tests/conftest.py
--- a/swh/provenance/tests/conftest.py
+++ b/swh/provenance/tests/conftest.py
@@ -13,6 +13,8 @@
 
 from swh.core.db import BaseDb
 from swh.journal.serializers import msgpack_ext_hook
+from swh.model.hashutil import hash_to_bytes
+from swh.model.model import Sha1Git
 from swh.model.tests.swh_model_data import TEST_OBJECTS
 from swh.provenance.postgresql.archive import ArchivePostgreSQL
 from swh.provenance.storage.archive import ArchiveStorage
@@ -114,13 +116,13 @@
 class SynthRelation(TypedDict):
     prefix: Optional[str]
     path: str
-    src: bytes
-    dst: bytes
+    src: Sha1Git
+    dst: Sha1Git
     rel_ts: float
 
 
 class SynthRevision(TypedDict):
-    sha1: bytes
+    sha1: Sha1Git
     date: float
     msg: str
     R_C: List[SynthRelation]
@@ -134,7 +136,7 @@
 
     Generated SynthRevision (typed dict) with the following elements:
 
-      "sha1": (bytes) sha1 of the revision,
+      "sha1": (Sha1Git) sha1 of the revision,
       "date": (float) timestamp of the revision,
       "msg": (str) commit message of the revision,
       "R_C": (list) new R---C relations added by this revision
@@ -144,8 +146,8 @@
     Each relation above is a SynthRelation typed dict with:
 
       "path": (str) location
-      "src": (bytes) sha1 of the source of the relation
-      "dst": (bytes) sha1 of the destination of the relation
+      "src": (Sha1Git) sha1 of the source of the relation
+      "dst": (Sha1Git) sha1 of the destination of the relation
       "rel_ts": (float) timestamp of the target of the relation
                 (related to the timestamp of the revision)
 
@@ -183,7 +185,7 @@
 def _mk_synth_rev(synth_rev) -> SynthRevision:
     assert synth_rev[0]["type"] == "R"
     rev = SynthRevision(
-        sha1=bytes.fromhex(synth_rev[0]["sha1"]),
+        sha1=hash_to_bytes(synth_rev[0]["sha1"]),
         date=float(synth_rev[0]["ts"]),
         msg=synth_rev[0]["revname"],
         R_C=[],
@@ -202,7 +204,7 @@
                     prefix=None,
                     path=row["path"],
                     src=rev["sha1"],
-                    dst=bytes.fromhex(row["sha1"]),
+                    dst=hash_to_bytes(row["sha1"]),
                     rel_ts=float(row["ts"]),
                 )
             )
@@ -214,7 +216,7 @@
                     prefix=None,
                     path=row["path"],
                     src=rev["sha1"],
-                    dst=bytes.fromhex(row["sha1"]),
+                    dst=hash_to_bytes(row["sha1"]),
                     rel_ts=float(row["ts"]),
                 )
             )
@@ -226,7 +228,7 @@
                     prefix=current_path,
                     path=row["path"],
                     src=rev["R_D"][-1]["dst"],
-                    dst=bytes.fromhex(row["sha1"]),
+                    dst=hash_to_bytes(row["sha1"]),
                     rel_ts=float(row["ts"]),
                 )
             )
diff --git a/swh/provenance/tests/data/generate_storage_from_git.py b/swh/provenance/tests/data/generate_storage_from_git.py
--- a/swh/provenance/tests/data/generate_storage_from_git.py
+++ b/swh/provenance/tests/data/generate_storage_from_git.py
@@ -11,6 +11,7 @@
 import yaml
 
 from swh.loader.git.from_disk import GitLoaderFromDisk
+from swh.model.hashutil import hash_to_bytes
 from swh.model.model import (
     Origin,
     OriginVisit,
@@ -90,7 +91,7 @@
                 # add a snapshot with branches from the input file
                 branches = {
                     f"refs/heads/{name}".encode(): SnapshotBranch(
-                        target=bytes.fromhex(all_branches[f"refs/heads/{name}"]),
+                        target=hash_to_bytes(all_branches[f"refs/heads/{name}"]),
                         target_type=TargetType.REVISION,
                     )
                     for name in visit["branches"]
diff --git a/swh/provenance/tests/test_origin_iterator.py b/swh/provenance/tests/test_origin_iterator.py
--- a/swh/provenance/tests/test_origin_iterator.py
+++ b/swh/provenance/tests/test_origin_iterator.py
@@ -6,9 +6,9 @@
 from swh.model.tests.swh_model_data import TEST_OBJECTS
 from swh.provenance.origin import CSVOriginIterator
 from swh.storage.algos.origin import (
-    iter_origins,
-    iter_origin_visits,
     iter_origin_visit_statuses,
+    iter_origin_visits,
+    iter_origins,
 )
 
 
@@ -21,9 +21,7 @@
                 swh_storage_with_objects, origin.url, visit.visit
             ):
                 if status.snapshot is not None:
-                    origins_csv.append(
-                        (status.origin, status.date.isoformat(), status.snapshot)
-                    )
+                    origins_csv.append((status.origin, status.snapshot))
     origins = list(CSVOriginIterator(origins_csv))
     assert origins
     assert len(origins) == len(
diff --git a/swh/provenance/tests/test_provenance_db.py b/swh/provenance/tests/test_provenance_db.py
--- a/swh/provenance/tests/test_provenance_db.py
+++ b/swh/provenance/tests/test_provenance_db.py
@@ -26,9 +26,7 @@
     archive = ArchiveStorage(swh_storage_with_objects)
     for status in TEST_OBJECTS["origin_visit_status"]:
         if status.snapshot is not None:
-            entry = OriginEntry(
-                url=status.origin, date=status.date, snapshot=status.snapshot
-            )
+            entry = OriginEntry(url=status.origin, snapshot=status.snapshot)
             origin_add(provenance, archive, [entry])
     # TODO: check some facts here
 
diff --git a/swh/provenance/tests/test_provenance_heuristics.py b/swh/provenance/tests/test_provenance_heuristics.py
--- a/swh/provenance/tests/test_provenance_heuristics.py
+++ b/swh/provenance/tests/test_provenance_heuristics.py
@@ -7,6 +7,7 @@
 
 import pytest
 
+from swh.model.hashutil import hash_to_bytes
 from swh.provenance.model import RevisionEntry
 from swh.provenance.revision import revision_add
 from swh.provenance.tests.conftest import (
@@ -82,7 +83,7 @@
     'cur' is a cursor to the provenance index DB.
     """
     if isinstance(sha1, str):
-        sha1 = bytes.fromhex(sha1)
+        sha1 = hash_to_bytes(sha1)
     cur.execute(f"SELECT date FROM {table} WHERE sha1=%s", (sha1,))
     return [date.timestamp() for (date,) in cur.fetchall()]
 
@@ -265,7 +266,7 @@
         db_occurrences = [
             (blob.hex(), rev.hex(), date.timestamp(), path.decode())
             for blob, rev, date, path in provenance.content_find_all(
-                bytes.fromhex(content_id)
+                hash_to_bytes(content_id)
             )
         ]
         if provenance.storage.with_path:
@@ -337,7 +338,7 @@
 
     for content_id, (rev_id, ts, paths) in expected_first.items():
         (r_sha1, r_rev_id, r_ts, r_path) = provenance.content_find_first(
-            bytes.fromhex(content_id)
+            hash_to_bytes(content_id)
         )
         assert r_sha1.hex() == content_id
         assert r_rev_id.hex() == rev_id