what do you mean by "the existing Group option" ? applications ?
Below "Applications" there is a "Group" combobox. But this was only a suggestion, I think implementing it like the "Applications" option would be fine too.
oooh. that one is completely different and does not impact item polling in any way 
it's sort of a method to add/delete/update item on many hosts at once.
Yes, I know 
. It was just a suggestion on the GUI implementation of this functionality. I mean, you have to be able to group items that you want polled together in some way using the web interface.
Related issue: ZBXNEXT-98.
I agree with your idea a "I believe it is essencial that the Zabbix server provides an option to poll different items at the same time, especially after calculated items were implemented." this feature is necessary for monitoring some items with more precise.
I hope this feature is to be added in next release.
Thanks
After some investigation on database structure of zabbix,I found that zabbix server,requests data according to "itemid" or such field in database.if it is true, I think with changing order of rows based on "itemid" we can sort order of requests ourself.
Is there any idea or suggestion about this please share.
I fully support this request.
But it should be considered in two aspects:
1. Simultaneity of the items polling.
2. Managed sequence to the items polling and trigger processing.
I will focus on the second aspect - I have not investigated in detail, but it seems i have noticed that a sequence of items polling depends on their ID.
It also seems sequence of the trigger processing, which use the same item, is also dependent on their ID.
At least this behaviour must be described in the documentation in the trigger dependencies point. And would really be nice to be able to manage it. See also ZBX-3163 for triggers.
Some forum discussion (in Russian): http://www.zabbix.com/forum/showthread.php?p=100140
Hi
After a few week since reporting this feature, now In our monitoring system I have to config zabbix so that fetches some items in order.
In last investigation on zabbix infrastructure, I have found that,zabbix fetch items based on "itemID".
My question is,Can I change itemID for few items?? if so,please guide me,which tables should I change also.
Thanks in advance
trigger evaluation might be ZBX-3163
I do miss this feature when making a graph of two items. One is the processor idle time and the other is processor active time. I have set them up in a stacked graph. When there is processor activity, the total won't be 100%, but is has gaps and peaks. It has nothing to do with triggers, only that items get retrieved at different times. When will there be a fix or is a fix in sight?
This would also solve ZBXNEXT-1049
This is such an important feature which would make a lot of things easier and faster.
This feature would be pretty much awesome.
Also, if this would do something like, pulling a swithport's items (all of them) together, that'd also save some resources when evaluating triggers, since the triggers using multiple items for a single port wouldn't have to be evaluated after every item's poll, just once, after an aggregated poll.
this probably wouldn't help that much with triggers, as the trigger parsing logic wouldn't differ much.
when you mention switch ports, i guess it's about snmp, which would also be different - instead, snmp bulkget would probably be what you want
ZBX-7649 solves the problem for ICMP ping items on the same host.
Not exactly what requested here, but it's possible in 3.0 using ZBXNEXT-2128 feature.
Just need to be careful to not overuse that feature to not have spikes of busy level for processes which gather information.
maybe a practical example helps - the cpu utilisation items, if queried at the same time, should sum up to 100%. with zabbix item polling distributed, it most often does not, and the top of the graph tends to have spikes over & dips under 100% - sometimes fairly large spikes & dips.
(1) an extension of this feature request - in addition to "affinity", there could also be "harmonisation".
for example, if we have one item collecting data during business hours, and another during off-hours, it would be nice to mark them for "harmonisation", so that the interval/poll time is aligned as much as possible.
if one of those items collects data every 30 minutes and the other every hour, the "harmonisation" would ensure that the interval between two values is always 30 minutes, and doesn't increase or decrease during the time when one item takes over the another.