diff --git a/swh/storage/checker/checker.py b/swh/storage/checker/checker.py
--- a/swh/storage/checker/checker.py
+++ b/swh/storage/checker/checker.py
@@ -34,7 +34,7 @@
             restore corrupted content.
     """
 
-    def __init__(self, config, root, depth, backup_urls):
+    def __init__(self, config, root, slicing, backup_urls):
         """ Create a checker that ensure the objstorage have no corrupted file.
 
         Args:
@@ -47,7 +47,7 @@
                 get a content.
         """
         self.config = config
-        self.objstorage = PathSlicingObjStorage(root, depth, slicing=2)
+        self.objstorage = PathSlicingObjStorage(root, slicing)
         self.backup_storages = [get_storage('remote_storage', [backup_url])
                                 for backup_url in backup_urls]
 
diff --git a/swh/storage/objstorage/api/server.py b/swh/storage/objstorage/api/server.py
--- a/swh/storage/objstorage/api/server.py
+++ b/swh/storage/objstorage/api/server.py
@@ -16,7 +16,7 @@
 
 DEFAULT_CONFIG = {
     'storage_base': ('str', '/tmp/swh-storage/objects/'),
-    'storage_depth': ('int', 3)
+    'storage_slicing': ('str', '0:2/2:4/4:6')
 }
 
 app = Flask(__name__)
@@ -31,8 +31,7 @@
 @app.before_request
 def before_request():
     g.objstorage = PathSlicingObjStorage(app.config['storage_base'],
-                                         app.config['storage_depth'],
-                                         slicing=2)
+                                         app.config['storage_slicing'])
 
 
 @app.route('/')
diff --git a/swh/storage/objstorage/objstorage.py b/swh/storage/objstorage/objstorage.py
--- a/swh/storage/objstorage/objstorage.py
+++ b/swh/storage/objstorage/objstorage.py
@@ -4,6 +4,10 @@
 # See top-level LICENSE file for more information
 
 
+ID_HASH_ALGO = 'sha1'
+ID_HASH_LENGTH = 40  # Size in bytes of the hash hexadecimal representation.
+
+
 class ObjStorage():
     """ High-level API to manipulate the Software Heritage object storage.
 
@@ -47,7 +51,7 @@
             "Implementations of ObjStorage must have a 'add' method"
         )
 
-    def restore(self, content, obj_id, *args, **kwargs):
+    def restore(self, content, obj_id=None, *args, **kwargs):
         """ Restore a content that have been corrupted.
 
         This function is identical to add_bytes but does not check if
diff --git a/swh/storage/objstorage/objstorage_pathslicing.py b/swh/storage/objstorage/objstorage_pathslicing.py
--- a/swh/storage/objstorage/objstorage_pathslicing.py
+++ b/swh/storage/objstorage/objstorage_pathslicing.py
@@ -12,12 +12,10 @@
 
 from swh.core import hashutil
 
-from .objstorage import ObjStorage
+from .objstorage import ObjStorage, ID_HASH_ALGO, ID_HASH_LENGTH
 from ..exc import ObjNotFoundError, Error
 
 
-ID_HASH_ALGO = 'sha1'
-
 GZIP_BUFSIZ = 1048576
 
 DIR_MODE = 0o755
@@ -84,33 +82,32 @@
     the value of the ID_HASH_ALGO constant (see hashutil for its meaning).
 
     To avoid directories that contain too many files, the object storage has a
-    given depth. Each depth level consumes a given amount of characters of
-    the object id.
+    given slicing. Each slicing correspond to a directory that is named
+    according to the hash of its content.
 
     So for instance a file with SHA1 34973274ccef6ab4dfaaf86599792fa9c3fe4689
     will be stored in the given object storages :
 
-    - depth=3, slicing=2 : 34/97/32/34973274ccef6ab4dfaaf86599792fa9c3fe4689
-    - depth=1, slicing=5 : 34973/34973274ccef6ab4dfaaf86599792fa9c3fe4689
+    - 0:2/2:4/4:6 : 34/97/32/34973274ccef6ab4dfaaf86599792fa9c3fe4689
+    - 0:1/0:5/    : 3/34973/34973274ccef6ab4dfaaf86599792fa9c3fe4689
 
     The files in the storage are stored in gzipped compressed format.
 
     Attributes:
         root (string): path to the root directory of the storage on the disk.
-        depth (int): number of subdirectories created to store a file.
-        slicing (int): number of hash character consumed for each
-            subdirectories.
+        bounds: list of tuples that indicates the beginning and the end of
+            each subdirectory for a content.
     """
 
-    def __init__(self, root, depth, slicing):
+    def __init__(self, root, slicing):
         """ Create an object to access a hash-slicing based object storage.
 
         Args:
             root (string): path to the root directory of the storage on
                 the disk.
-            depth (int): number of subdirectories created to store a file.
-            slicing (int): number of hash character consumed for each
-                subdirectories.
+            slicing (string): string that indicates the slicing to perform
+                on the hash of the content to know the path where it should
+                be stored.
         """
         if not os.path.isdir(root):
             raise ValueError(
@@ -118,8 +115,20 @@
             )
 
         self.root = root
-        self.depth = depth
-        self.slicing = slicing
+        # Make a list of tuples where each tuple contains the beginning
+        # and the end of each slicing.
+        self.bounds = [
+            slice(*map(int, sbounds.split(':')))
+            for sbounds in slicing.split('/')
+            if sbounds
+        ]
+
+        max_endchar = max(map(lambda bound: bound.stop, self.bounds))
+        if ID_HASH_LENGTH < max_endchar:
+            raise ValueError(
+                'Algorithm %s has too short hash for slicing to char %d'
+                % (ID_HASH_ALGO, max_endchar)
+            )
 
     def __contains__(self, obj_id):
         """ Check whether the given object is present in the storage or not.
@@ -173,20 +182,8 @@
         Returns:
             Path to the directory that contains the required object.
         """
-        if len(hex_obj_id) < self.depth * self.slicing:
-            raise ValueError(
-                'Object id "%s" is to short for %d-slicing at depth %d'
-                % (hex_obj_id, self.slicing, self.depth)
-            )
-
-        # Compute [depth] substrings of [hex_obj_id], each of length [slicing],
-        # starting from the beginning.
-        id_steps = [hex_obj_id[i * self.slicing:
-                               i * self.slicing + self.slicing]
-                    for i in range(self.depth)]
-        steps = [self.root] + id_steps
-
-        return os.path.join(*steps)
+        slices = [hex_obj_id[bound] for bound in self.bounds]
+        return os.path.join(self.root, *slices)
 
     def _obj_path(self, hex_obj_id):
         """ Compute the full path to an object into the current storage.
@@ -331,7 +328,7 @@
                 a tuple (batch size, batch).
             """
             dirs = []
-            for level in range(self.depth):
+            for level in range(len(self.bounds)):
                 path = os.path.join(self.root, *dirs)
                 dir_list = next(os.walk(path))[1]
                 if 'tmp' in dir_list:
diff --git a/swh/storage/storage.py b/swh/storage/storage.py
--- a/swh/storage/storage.py
+++ b/swh/storage/storage.py
@@ -68,7 +68,9 @@
         except psycopg2.OperationalError as e:
             raise StorageDBError(e)
 
-        self.objstorage = PathSlicingObjStorage(obj_root, depth=3, slicing=2)
+        # TODO this needs to be configured
+        self.objstorage = PathSlicingObjStorage(obj_root,
+                                                slicing='0:2/2:4/4:6')
 
     def content_add(self, content):
         """Add content blobs to the storage
diff --git a/swh/storage/tests/test_archiver.py b/swh/storage/tests/test_archiver.py
--- a/swh/storage/tests/test_archiver.py
+++ b/swh/storage/tests/test_archiver.py
@@ -37,7 +37,7 @@
         # Launch the backup server
         self.backup_objroot = tempfile.mkdtemp(prefix='remote')
         self.config = {'storage_base': self.backup_objroot,
-                       'storage_depth': 3}
+                       'storage_slicing': '0:2/2:4/4:6'}
         self.app = app
         super().setUp()
 
diff --git a/swh/storage/tests/test_checker.py b/swh/storage/tests/test_checker.py
--- a/swh/storage/tests/test_checker.py
+++ b/swh/storage/tests/test_checker.py
@@ -43,8 +43,8 @@
         # Connect to an objstorage
         config = {'batch_size': 10}
         path = tempfile.mkdtemp()
-        depth = 3
-        self.checker = ContentChecker(config, path, depth, 'http://None')
+        slicing = '0:2/2:4/4:6'
+        self.checker = ContentChecker(config, path, slicing, 'http://None')
         self.checker.backup_storages = [MockBackupStorage(),
                                         MockBackupStorage()]
 
diff --git a/swh/storage/tests/test_objstorage_api.py b/swh/storage/tests/test_objstorage_api.py
--- a/swh/storage/tests/test_objstorage_api.py
+++ b/swh/storage/tests/test_objstorage_api.py
@@ -3,7 +3,6 @@
 # License: GNU General Public License version 3, or any later version
 # See top-level LICENSE file for more information
 
-import os
 import tempfile
 import unittest
 
@@ -24,7 +23,7 @@
 
     def setUp(self):
         self.config = {'storage_base': tempfile.mkdtemp(),
-                       'storage_depth': 3}
+                       'storage_slicing': '0:1/0:5'}
         self.app = app
         super().setUp()
         self.objstorage = RemoteObjStorage(self.url())
@@ -65,20 +64,12 @@
     @istest
     def content_check_invalid(self):
         content = bytes('content_check_invalid', 'utf8')
-        id = self.objstorage.content_add(content)
-        hex_obj_id = hashutil.hash_to_hex(id)
-        dir_path = os.path.join(
-            self.config['storage_base'],
-            *[hex_obj_id[i*2:i*2+2]
-              for i in range(int(self.config['storage_depth']))]
-        )
-        path = os.path.join(dir_path, hex_obj_id)
-        content = list(content)
-        with open(path, 'bw') as f:
-            content[0] = (content[0] + 1) % 128
-            f.write(bytes(content))
+        invalid_id = hashutil.hashdata(b'invalid content')['sha1']
+        # Add the content with an invalid id.
+        self.objstorage.content_add(content, invalid_id)
+        # Then check it and expect an error.
         with self.assertRaises(Error):
-            self.objstorage.content_check(id)
+            self.objstorage.content_check(invalid_id)
 
     @istest
     def content_check_valid(self):
diff --git a/swh/storage/tests/test_objstorage_pathslicing.py b/swh/storage/tests/test_objstorage_pathslicing.py
--- a/swh/storage/tests/test_objstorage_pathslicing.py
+++ b/swh/storage/tests/test_objstorage_pathslicing.py
@@ -19,11 +19,9 @@
 
     def setUp(self):
         super().setUp()
-        self.depth = 3
-        self.slicing = 2
+        self.slicing = '0:2/2:4/4:6'
         self.tmpdir = tempfile.mkdtemp()
-        self.storage = PathSlicingObjStorage(self.tmpdir, self.depth,
-                                             self.slicing)
+        self.storage = PathSlicingObjStorage(self.tmpdir, self.slicing)
 
     def content_path(self, obj_id):
         hex_obj_id = hashutil.hash_to_hex(obj_id)