I agree but there are more factors then just those as far as I have discover and I poll 5000 device per data center about 1600 devices per poller per poller 60000 oid ever 5 mins it take 16 seconds for zenperf to run but about 4mins to write data
on here is a list of factors and explanations number of device -- this is only a factor when reading the config when you have many devices say 2000 it has to read each config threw the zenhub number of oid -- this is a factor because the more oids you pull the greater number of files zenoss have to open raid configurations -- this is a factor because of the number of reads and writes number of drives -- the more spindles zenoss has the more it can write in a shorter time frame speed of drives -- 10k vs 15k drives is a major boost memory -- if you have extra zenoss will use it to store the data in memory instead or reading the values from file each time cpu -- this is only a factor in zenhub on the main box when you have one hub trying to talk to all the daemons it bogs it down so to have multiple hubs in which each hub has its own cpu is the best. amount of data stores -- this is a factor because the greater number of lines in your rrd the bigger your rrd file size this cause longer read times. how often you poll for data -- this is a factor to give the hard drives enough time to write the data. I wish there was a magic formula to calculate what you need but haven't found it yet! sorry! hope this helps, Pete Charles Byam wrote: > On 1/15/09 10:18 AM, "charlesdf23" <[email protected]> wrote: > > >> I would like to start out stating that I am very impressed with this product >> and we going through a proof of concept/ eval. >> >> I'm seeing drastic performance issues with my fresh Zenoss install. >> >> I have 493 devices (mostly cisco switches) and according the the zenperfsnmp >> log it's taking around 500 seconds to complete. >> >> I also receive numerous >> Timeout connecting to zenhub: is it running? >> >> and for some reason zenwin and zeneventlog keep stopping themselves. >> >> This is a 2x quad core box with 8gb of memory in it. >> >> Any ideas or thoughts to hopefully troubleshoot this? >> > > It's not necessarily how many devices your poling, but how many data points > your collecting. I've found that about 80k data points is the max any one > server can handle before response time becomes an issue. Zenoss is very > write intensive and anything you can do to improve performance of the disk > sub-system will directly improve server performance. > > To check how many data points your collecting: > > grep OID `find $ZENHOME/log -name zenperfsnmp.log` | tail > > Top will give you an idea of how much I/O wait time your dealing with. > > On my HP Proliant DL360 G5's I'm using RAID 1+0 and allocated 75% of the > controller cache to writes. > > You could also create another logical drive to store your data on > ($ZENHOME/perf) and set the mount point option to noatime. > > Hope this helps. > -- > Chuck > > > > > -- ------------------------------------------------------ Peter Mistich Linux System Engineer Rackspace Managed Hosting (210)312-4821 ------------------------------------------------------ Confidentiality Notice: This e-mail message (including any attached or embedded documents) is intended for the exclusive and confidential use of the individual or entity to which this message is addressed, and unless otherwise expressly indicated, is confidential and privileged information of Rackspace. Any dissemination, distribution or copying of the enclosed material is prohibited. If you receive this transmission in error, please notify us immediately by e-mail at [email protected], and delete the original message. Your cooperation is appreciated. _______________________________________________ zenoss-users mailing list [email protected] http://lists.zenoss.org/mailman/listinfo/zenoss-users
