diff --git a/swh/objstorage/__init__.py b/swh/objstorage/__init__.py
--- a/swh/objstorage/__init__.py
+++ b/swh/objstorage/__init__.py
@@ -11,6 +11,8 @@
     MultiplexerObjStorage, StripingObjStorage)
 from swh.objstorage.multiplexer.filter import add_filters
 from swh.objstorage.backends.seaweed import WeedObjStorage
+from swh.objstorage.backends.generator import RandomGeneratorObjStorage
+
 
 __all__ = ['get_objstorage', 'ObjStorage']
 
@@ -20,6 +22,7 @@
     'remote': RemoteObjStorage,
     'memory': InMemoryObjStorage,
     'weed': WeedObjStorage,
+    'random': RandomGeneratorObjStorage,
 }
 
 _STORAGE_CLASSES_MISSING = {
diff --git a/swh/objstorage/backends/generator.py b/swh/objstorage/backends/generator.py
new file mode 100644
--- /dev/null
+++ b/swh/objstorage/backends/generator.py
@@ -0,0 +1,126 @@
+from itertools import count, islice
+import random
+import io
+import functools
+import logging
+
+from swh.objstorage.objstorage import (
+    ObjStorage, DEFAULT_CHUNK_SIZE, DEFAULT_LIMIT)
+
+logger = logging.getLogger(__name__)
+
+
+class Randomizer:
+    def __init__(self):
+        self.size = 0
+        self.data = None
+
+    def read(self, size):
+        if size > self.size:
+            with open('/dev/urandom', 'rb') as fobj:
+                self.data = fobj.read(2*size)
+                self.size = len(self.data)
+        idx = random.randint(0, self.size - size - 1)
+        return self.data[idx:idx+size]
+
+
+def gen_filesize(total=None, filesize=None):
+    '''generates numbers according to the rought distribution of file size in the
+    SWH archive
+    '''
+    # these are the histogram bounds of the pg content.length column
+    bounds = [0, 2, 72, 119, 165, 208, 256, 300, 345, 383, 429, 474, 521, 572,
+              618, 676, 726, 779, 830, 879, 931, 992, 1054, 1119, 1183, 1244,
+              1302, 1370, 1437, 1504, 1576, 1652, 1725, 1806, 1883, 1968, 2045,
+              2133, 2236, 2338, 2433, 2552, 2659, 2774, 2905, 3049, 3190, 3322,
+              3489, 3667, 3834, 4013, 4217, 4361, 4562, 4779, 5008, 5233, 5502,
+              5788, 6088, 6396, 6728, 7094, 7457, 7835, 8244, 8758, 9233, 9757,
+              10313, 10981, 11693, 12391, 13237, 14048, 14932, 15846, 16842,
+              18051, 19487, 20949, 22595, 24337, 26590, 28840, 31604, 34653,
+              37982, 41964, 46260, 51808, 58561, 66584, 78645, 95743, 122883,
+              167016, 236108, 421057, 1047367, 55056238]
+
+    nbounds = len(bounds)
+    for i in count():
+        if total and i > total:
+            break
+        if filesize is not None:
+            yield filesize
+        else:
+            idx = random.randint(1, nbounds-1)
+            lower = bounds[idx-1]
+            upper = bounds[idx]
+            yield random.randint(lower, upper-1)
+
+
+def gen_random_file(total=None, filesize=None):
+    '''generates random file content which sizes roughly follows the SWH
+    archive file size distribution
+
+    '''
+    randomizer = Randomizer()
+    for objsize in gen_filesize(total, filesize):
+        yield randomizer.read(objsize)
+
+
+class RandomGeneratorObjStorage(ObjStorage):
+    '''A stupid read-only storage that generated blobs for testing purpose.
+    '''
+
+    def __init__(self, filesize=None, total=None, **kwargs):
+        super().__init__()
+        if filesize:
+            filesize = int(filesize)
+        self.filesize = filesize
+        if total:
+            total = int(total)
+        self.total = total
+        self._gen = None
+
+    @property
+    def gen(self):
+        if self._gen is None:
+            self._gen = gen_random_file(self.total, self.filesize)
+        return self._gen
+
+    def check_config(self, *, check_write):
+        return True
+
+    def __contains__(self, obj_id, *args, **kwargs):
+        return False
+
+    def __iter__(self):
+        i = 0
+        while True:
+            j = yield (b'%d' % i)
+            if self.total and i > self.total:
+                logger.debug('DONE')
+                break
+            if j is not None:
+                i = j
+            else:
+                i += 1
+
+    def get(self, obj_id, *args, **kwargs):
+        return next(self.gen)
+
+    def add(self, content, obj_id=None, check_presence=True, *args, **kwargs):
+        pass
+
+    def check(self, obj_id, *args, **kwargs):
+        return True
+
+    def delete(self, obj_id, *args, **kwargs):
+        return True
+
+    def get_stream(self, obj_id, chunk_size=DEFAULT_CHUNK_SIZE):
+        data = io.BytesIO(next(self.gen))
+        reader = functools.partial(data.read, chunk_size)
+        yield from iter(reader, b'')
+
+    def list_content(self, last_obj_id=None, limit=DEFAULT_LIMIT):
+        it = iter(self)
+        if last_obj_id:
+            next(it)
+            it.send(int(last_obj_id))
+        return islice(it, limit)