Dear Wiki user, You have subscribed to a wiki page or wiki category on "Hadoop Wiki" for change notification.
The "ZooKeeper/ServiceLatencyOverview" page has been changed by PatrickHunt. http://wiki.apache.org/hadoop/ZooKeeper/ServiceLatencyOverview?action=diff&rev1=11&rev2=12 -------------------------------------------------- All systems had 16gig memory available, however unless specifically noted the JVM's -Xmx option was used to limit the size of the JVM heap to 512m. - All systems had 7200RPM SATA drives. + All systems had 7200RPM SATA drives. hdparm reported: + * Timing cached reads: 22980 MB in 1.99 seconds = 11532.40 MB/sec + * Timing buffered disk reads: 266 MB in 3.01 seconds = 88.29 MB/sec All systems are interconnected with 1gig ethernet, approximate ping time is steady at 0.09ms roundtrip. @@ -45, +47 @@ * zk-latencies.py from http://github.com/phunt/zk-smoketest was used as the test client. * python 2.4.3 + == Results == + + the results are as follows for standalone and 3 node ensemble + + Note: in each case below I give only a single client's results, regardless of the number of clients run in the tests. I do this to limit the length of this document, the results that I do show are for the last, longest running client, so in general these are the worst performing results of any client in the test. I also verify that all clients completed successfully and without warnings or errors in the client log. + + == Results for Standalone Server == + + The Standalone version of ZooKeeper was run on a single host ("host60"). + - The client was run as follows unless otherwise noted: + The client(s) was run as follows unless otherwise noted: {{{ - zk-latencies.py --servers "host60:2181" --root_znode=/`hostname` --znode_size=100 --znode_count=10000 --timeout=5000 --watch_multiple=5 + zk-latencies.py --cluster "host60:2181" --root_znode=/`hostname` --znode_size=100 --znode_count=10000 --timeout=5000 --watch_multiple=5 }}} This means each client: @@ -57, +69 @@ * sets 5 watches on each of the znodes it creates, for 50k znodes by each client As a result, each client will send ~110,000 requests to the server during each test. - - == Results == - - the results are as follows for standalone and 3 node ensemble - - Note: in each case below I give only a single client's results, regardless of the number of clients run in the tests. I do this to limit the length of this document, the results that I do show are for the last, longest running client, so in general these are the worst performing results of any client in the test. I also verify that all clients completed successfully and without warnings or errors in the client log. - - == Results for Standalone Server == === 1 Core - 1 Client === @@ -237, +241 @@ host65: 0inputs+0outputs (0major+75539minor)pagefaults 0swaps }}} + == Results for 3 Server Ensemble == + + A 3 Server Ensemble (3 distinct hosts) was used for this test ("host60", "host62" & "host63"). + + The client was run as follows unless otherwise noted: + + {{{ + zk-latencies.py --cluster "host60:2181,host61:2181,host63:2181" --root_znode=/`hostname` --znode_size=100 --znode_count=10000 --timeout=5000 --watch_multiple=5 + }}} + + This means each client: + * has a 5 second timeout + * creates 10k znodes of size 100 bytes + * sets 5 watches on each of the znodes it creates, for 50k znodes by each client + + As a result, each client will send ~110,000 requests to the server during each test. + + === 1 Core - 1 Client === + + 110k operations against the server with elapsed time of 18 seconds + + {{{ + }}} + + === 1 Core - 10 Client === + + 1.1 million operations against the server with elapsed time of 58 seconds + + {{{ + }}} + + === 1 Core - 20 Client === + + 2.2 million operations against the server with elapsed time of 150 seconds + + {{{ + }}} + + === 2 Core - 1 Client === + + 110k operations against the server with elapsed time of 20 seconds + + {{{ + }}} + + === 2 Core - 10 Client === + + 1.1 million operations against the server with elapsed time of 53 seconds + + {{{ + }}} + + === 2 Core - 20 Client === + + 2.2 million operations against the server with elapsed time of 108 seconds + + {{{ + }}} + + === 4 Core - 1 Client === + + 110k operations against the server with elapsed time of 20 seconds + + {{{ + }}} + + === 4 Core - 10 Client === + + 1.1 million operations against the server with elapsed time of 43 seconds + + {{{ + }}} + + === 4 Core - 20 Client === + + 2.2 million operations against the server with elapsed time of 95 seconds + + {{{ + }}} +
