X-Git-Url: http://git.vrable.net/?a=blobdiff_plain;f=cloudbench%2Fparalleltest.py;h=3f74b2fa8b050c8ac9e171e64334991c03f47059;hb=432c0ffa8e2437880a4ed593819881df8c1eaac3;hp=5084b62229e734d127a14f07b9c035e166ce8223;hpb=b49973e78300a264aa8e3b985aa125e0e0e89d8c;p=bluesky.git

diff --git a/cloudbench/paralleltest.py b/cloudbench/paralleltest.py
index 5084b62..3f74b2f 100755
--- a/cloudbench/paralleltest.py
+++ b/cloudbench/paralleltest.py
@@ -10,7 +10,7 @@ import sys, threading, time, Queue
 import azure
 
 BUCKET_NAME = 'mvrable-benchmark'
-SIZES = [64, 4096, 32 << 10, 256 << 10, 1 << 20, 4 << 20, 32 << 20]
+SIZES = [(1 << s) for s in range(12, 23)]
 
 class S3TestConnection:
     def __init__(self):
@@ -48,16 +48,86 @@ def parallel_get(name, connections, delay1=0.0):
     res = []
     while not q.empty():
         res.append(q.get())
-    return res
 
-def run_test(size, threads, num):
+    if len(res) == len(connections):
+        return res
+
+def parallel_multiget(names, connections, repeat=1):
+    requests = Queue.Queue()
+    results = [[threading.Lock(), None] for n in names]
+    for i in range(len(names)):
+        for _ in range(repeat):
+            requests.put((names[i], results[i]))
+
+    threads = []
+    def launcher(c, requests):
+        while True:
+            try:
+                (n, r) = requests.get(block=False)
+                # Possible data race here but it should be harmless
+                if r[1] is None:
+                    res = c.get_object(n)
+                    r[0].acquire()
+                    if r[1] is None: r[1] = time.time()
+                    r[0].release()
+                requests.task_done()
+            except Queue.Empty:
+                return
+    for i in range(len(connections)):
+        c = connections[i]
+        threads.append(threading.Thread(target=launcher, args=(c, requests)))
+    start_time = time.time()
+    for i in range(len(threads)):
+        threads[i].start()
+    requests.join()
+
+    return max(x[1] for x in results) - start_time
+
+def run_test(size, threads, num, logfile=sys.stdout, delay=1.0):
     connections = [S3TestConnection() for _ in range(threads)]
     for i in range(num):
+        print "    ...test", i
         res = parallel_get('file-%d-%d' % (size, i), connections)
-        print res
-        time.sleep(1.0)
+        if res is not None:
+            logfile.write(str(min(res)) + "\n")
+        if delay > 0:
+            time.sleep(delay)
+
+connections = [S3TestConnection() for _ in range(128)]
+logfile = open('multifetch-simulation.data', 'a')
+for s in [(1 << s) for s in range(16, 27)]:
+    print "Priming objects: %d-byte objects" % (s,)
+    run_test(s, 1, 100, open('/dev/null', 'w'), 0.0)
+
+    for blocksize in [x << 20 for x in (4, 8, 16, 32, 64, 128)]:
+        if s > blocksize: continue
+        for t in [4, 2, 1]:
+            for rep in range(10):
+                count = blocksize // s
+                print "Running tests: %d-byte blocks, %d-byte objects, %d parallel fetches" % (blocksize, s, t)
+                print "Object count:", count
+                if count * t > len(connections):
+                    conns = connections
+                else:
+                    conns = connections[0 : count * t]
+
+                objects = ['file-%d-%d' % (s, i % 100) for i in range(count)]
+                r = parallel_multiget(objects, conns, t)
+                print r
+                logfile.write('%s\t%s\t%s\t%s\t%s\n' % (s, blocksize >> 20, t, len(conns), r))
+                logfile.flush()
+                time.sleep(2.0)
+sys.exit(0)
+
+for s in SIZES:
+    print "Priming objects: %d-byte objects" % (s,)
+    logfile = open('/dev/null', 'w')
+    run_test(s, 1, 100, logfile, 0.0)
 
-run_test(32768, 4, 500)
+    for t in [4, 2, 1]:
+        print "Running tests: %d-byte objects, %d parallel fetches" % (s, t)
+        logfile = open('parallel-%d-%d.data' % (s, t), 'w')
+        run_test(s, t, 100, logfile)
 sys.exit(0)
 
 if __name__ == '__main__':