http://www.datastax.com/support-forums/tags/read Software, Support, and Training for Apache Cassandra en-US Tue, 21 May 2013 20:38:49 +0000 http://bbpress.org/?v=1.0.3 q http://www.datastax.com/support-forums/search.php http://www.datastax.com/support-forums/topic/cassandra-read-write-performance-matrix#post-2241 Wed, 20 Jun 2012 13:03:05 +0000 Anonymous 2241@http://www.datastax.com/support-forums/ <p>I am doing a comparison test between Cassandra and MS SQL. I have only one table/column family with 10 columns in it. I am trying to calculate the transaction per second Cassandra and MS SQL execute. While writing the data Cassandra process 5000 - 7000 transactions per second and MS SQL 70 - 90 transactions per second.</p> <p>While reading the data Cassandra only process 500 - 700 transactions per second and MS SQL 3000 - 6000 transactions per second which is unexpected for me. I was expecting Cassandra to be faster than MS SQL both in case of reading and writing.</p> <p>In case of Cassandra, I have 3 nodes cluster running remotly and MS SQL also running remotly. You can find the code below. I am using the same code for MS SQL and Cassandra</p> <p>for(int i=0;i&lt;BenchmarkConfig.insertData;i++)</p> <p>{</p> <p>orderId = Generate Radom ID;</p> <p>rs = statement.executeQuery("Select * From ABC where KEY = "+orderId);</p> <p>}</p> <p>Can you please help me out why reading is that much slow in Cassandra ? To save the time to investigate I am using Random Partitioner for equal distribution and incase of reading, I have not specified any consistency level by default which is 1. </p> http://www.datastax.com/support-forums/topic/stress-tool-cant-read#post-1408 Wed, 21 Mar 2012 10:02:25 +0000 albertnices 1408@http://www.datastax.com/support-forums/ <p>If you have lots of problem from stress just take meditation 5-10 min after that you feel very well. </p> http://www.datastax.com/support-forums/topic/stress-tool-cant-read#post-1323 Wed, 07 Mar 2012 13:45:15 +0000 Anonymous 1323@http://www.datastax.com/support-forums/ <p>Did you try -o read -n 100000? The issue isn't that you were trying to read with a different n, it's that you were trying to read with an n which has a different order of magnitude. When trying to read -n 1, see that it was trying to read key 0? That key doesn't exist from a -n 100000 insert, since it will have only inserted keys 000000-099999.</p> <p>To explain it more generally and without so many confusing sets of zeroes...<br /> If you go into Data Explorer in OpsCenter, or simply do a list CF in cassandra-cli, you'll see that each key generated by a -o insert -n x will have the same number of digits as x does. This is the problem you're running into when trying to -o read -n y. If y doesn't have the same number of digits as x, the read will fail as it's trying to access keys that don't exist.</p> <p>In general, you'll want to only use the stress -o read with the same -n you used during the insert. The farthest you can deviate from this would be to -o insert -n 9999 and then -o read -n 1000 (And similar 9...9-&gt;100...0 patterns), since they have the same number of digits they will both be accessing the same keys. </p> http://www.datastax.com/support-forums/topic/stress-tool-cant-read#post-1319 Wed, 07 Mar 2012 09:54:57 +0000 Anonymous 1319@http://www.datastax.com/support-forums/ <p>Well I did not know that but even with a<br /> stress -d 127.0.0.1 -o read -n 1 it does not work:</p> <p>total,interval_op_rate,interval_key_rate,avg_latency,elapsed_time<br /> Operation [0] retried 10 times - error reading key 0</p> <p>0,0,0,NaN,0<br /> END</p> <p>I'm sure data are in the DB.</p> <p>thx </p> http://www.datastax.com/support-forums/topic/stress-tool-cant-read#post-1304 Tue, 06 Mar 2012 13:43:42 +0000 Anonymous 1304@http://www.datastax.com/support-forums/ <p>stress -d 127.0.0.1 -n 100000<br /> will write rows with key values 000000-099999, when you then do<br /> stress -d 127.0.0.1 -o read<br /> without a -n, it is defaulting to -n 1000000, which will try to read keys 0000000-0999999 (notice the extra digit). Switch your stress read to:<br /> stress -d 127.0.0.1 -o read -n 100000<br /> and see if that works for you. </p> http://www.datastax.com/support-forums/topic/stress-tool-cant-read#post-1302 Tue, 06 Mar 2012 08:33:34 +0000 Anonymous 1302@http://www.datastax.com/support-forums/ <p>Hello I have a problem with the stress tool.<br /> I've played this:<br /> stress -d 127.0.0.1 -n 100000<br /> no problem<br /> then I do stress -d 127.0.0.1 -o read<br /> and here I have many error messages like this one:<br /> Operation [15] retried 10 times - error reading key 0526289<br /> and at the end:<br /> 0,0,0,NaN,2</p> <p>I'm running cassandra 1.0.8 under debian suqeeze and java 1.6.</p> <p>Does it should work?</p> <p>thx </p> http://www.datastax.com/support-forums/topic/cassandra-read-write-performance-matrix#post-1094 Tue, 07 Feb 2012 17:39:28 +0000 thobbs 1094@http://www.datastax.com/support-forums/ <p>What client are you using to do the read? What are you reading (one row, all 100 rows)? How are you measuring how long it takes? </p> http://www.datastax.com/support-forums/topic/cassandra-read-write-performance-matrix#post-1092 Tue, 07 Feb 2012 11:43:41 +0000 Anonymous 1092@http://www.datastax.com/support-forums/ <p>hi all,</p> <p>i am new to Cassandra<br /> currently i am configured Cassandra 0.8.6 with single node in a unix server.<br /> setting Cassandra JVM memory 512 and max 1024 and remaining as usual default settings.</p> <p>i had inserted some sample data with 3 columns around 100 rows.<br /> i checked that it takes around 2900 milliseconds for reading.</p> <p>can any one tell what should be the read and write time with a single node cluster that Cassandra Specs says.<br /> As in other RDBMS idle read and write should not take more than 50 milli seconds.</p> <p>what are the things in need to cross check to improve my Cassandra Performance?</p> <p>we are currently evaluating Cassandra for our data storage.</p> <p>Thanks in advance </p>