I'd like to measure the read query capacity of an existing cluster using the Monitoring History Tool.
My objective is to determine the average "Read Queries Per Second" during a spurt of query activity. My plan is to use the "Query Read Rate" from the Disk I/O graph. Here's the info: http://docs.marklogic.com/guide/monitoring/history#id_21175 I ran some JMeter scripts to measure the average request response time which was .74 seconds. During this time, I saw the following Disk I/O and Memory I/O metrics. Disk I/O Query Read Rate 00.03 MB/sec Merge Read Rate 28.00 MB/sec Memory I/O Page-In Rate: 22,436.3 pages/sec Page-Out Rate: 19,345.8 pages/sec How does this translate to Queries Per Second (QPS)? The average document (fragment) size is 4 KB (4096 bytes). The Query Read Rate is .03 MB/sec (30720 bytes/sec) Since each query returns a 4K document then 30720 divided by 4096 is 7.5 QPS. If each query returned just a 1024 byte search snippet then it would be 30 QPS. So I assume my answer QPS rates are: - 7.5 QPS when returning documents - 30.0 QPS when returning search snippets Please let me know if I'm missing something. Gary Russo NoSQL Architect and Developer http://garyrusso.wordpress.com http://twitter.com/garyprusso _______________________________________________ General mailing list [email protected] Manage your subscription at: http://developer.marklogic.com/mailman/listinfo/general
