RPM Clients

RPM was designed to provide an ultra-lightweight and ultra-fast means of determining the busiest processes and processors in large collections of NonStop servers.  Unlike other performance monitors such as Measure, ASAP, or Offender; the RPM product monitors processes without requiring Measure and without requiring any disk I/O whatsoever. The result is that RPM is extremely lightweight. RPM also provide super-scalability, eg it can monitor from 1 Cpu up to 1024+ processors. RPM easily monitors literally millions of processes each second, quickly determining which of these are the busiest and providing displays such as the ones shown on these web pages. Because RPM is memory based, it provides the absolute fastest way of determined the busiest processes and processors. In concert with this, RPM provides statistics via ultra-light device support, eg support for VT100, ANSI, T6530, and TTY mode devices.

However because of the wide utility of RPM, there are also advanced thin-web and rich fat-workstation clients that allow RPM statistics to be analyzed and displayed in graphical form, again without any disk I/O.   Below are examples of beta versions of RPM fat and thin clients. 

RPM Thin-Client

Below is an example of the RPM thin-client display.  It consists of a scoreboard that summarizes the total number of busiest Cpus and busiest Processes displayed, and the state of each of these.  This "scoreboard" is followed by grid-graph tabular displays that show the values of a variety of performance statistics (this is just an example, RPM is capable of displaying additional entity statistics).

NodeName	Cpu	hh:mm:ss	Busy%		QL		Disk	Chits		Mem	Used		Pcb
\ALABSYS	00	14:30	17		0		14	77		4294	13.23		99
\ALABSYS	01	14:30	13		0		0	0		4294	12.01		110
\ALABSYS	02	14:30	95		1		108	45		4294	7.62		73
\ALABSYS	03	14:30	2		0		0	0		4294	5.57		72
\SANFR	00	14:30	0		0		0	9		4294	17.51		27
\SANFR	01	14:30	0		0		0	0		4294	15.97		43
\SANFR	02	14:30	0		0		0	0		4294	10.48		23
\SANFR	03	14:30	0		0		0	0		4294	10.33		26
\TREE	00	14:30	2		0		0	63		2147	13.02		43
\TREE	01	14:30	1		0		0	0		2147	12.38		65

NodeName	hh:mm:ss	Busy%		PName	Program Object Filename	PRI	Userid	Pages
\ALABSYS	14:30	13.14		$SPIN	$Data.Mmspin.Spin	1	66,1	58
\ALABSYS	14:30	6.35		$ZXYZ	$Data.Data.Driver	160	255,255	91
\ALABSYS	14:30	93		$DB	$Data.Server.Database	160	255,34	95
\ALABSYS	14:30	.96		$MON	$System.Sys03.Osimage	201	255,255	55
\SANFR	14:30	.01		$MON	$System.Sys00.Nmontor	201	255,255	116
\SANFR	14:30	.06		$RPMX	$System.System.Driver	169	255,255	66
\SANFR	14:30	.01		$MON	$System.Sys00.Nmontor	201	255,255	83
\SANFR	14:30	.01		$MON	$System.Sys00.Nmontor	201	255,255	78
\TREE	14:30	.31		$TMP	$System.Sys00.Tmftmp	204	255,255	153
\TREE	14:30	.09		$MON	$System.Sys00.Osimage	201	255,255	80

RPM Fat-Client

Below is an example of the RPM fat-client display.  It consists of a rich interactive graph-grid displays that allow sorting, filtering, selection of the busiest Cpus and busiest Processes. State filtering is also possible on each of these.  Grid-graph tabular displays can be interactively modified to allow changes to types of performance statistics displayed, including interactive changing of color-coded thresholds (below is just an example, RPM is capable of displaying additional entity statistics).

Below is another example of the RPM fat-client display history (shows Cpu utilization over a 5 minute period).  It shows the history of Cpu 00 utilization.  RPM clearly shows process $SPIN is progressively busy, forming a saw-tooth graph of utilization.

An "early adopter" version of RPM fat/thin clients are available to interested selected RPM customers with networks consisting of 100 NonStop Cpus or greater. For additional information, please contact Support@NonStopRPM.Com,

Questions or comments:
Support@NonStopRPM.Com
Page modified: July 3, 2009