Archived Trends

[ziadrached]

Hello,
We are faced with the problem of reading Archived Trends . The system takes too much time and resources to read trends from a certain archive
The following readings are observed
a 5 days relative interval of 4 archived values the system would take about 4 seconds
a 10 days relative interval of the same 4 archived values the system would take about 15 seconds
the problem is significant when the interval is higher than 20 days for example
a 30 days relative interval of the same 4 archived values would take more than 5 minutes and sometimes halt the system for a long time ,where a forced system shutdown would be the only remedy.
And the required archive reading is at least 1 year. so imagine the time needed for reading values at such an interval.
note that when reading a 30 days interval we observed that the process zenonrt.exe process would consume a memory more than 1.7Gb

The PC is a Core 2 Duo 2.53 with 2 Gb of RAM
The Zenon Version is 6.22 SP0 build0 upgraded to Build6
The archive has a scan rate of 10 seconds and saved every 15 min in the default arx format

Is there a remedy for such a problem?
Best Regards.

[markclemens]

You can use following archives to achieve data reduction and consolidation of your data.

Please see the online help manual and tutorial for more information on this topic.

Viewing historical data of e.g. the past 30 days, for one variable, would result in 259.200 (10 seconds) values. If you view this amount of data in a trend which has a width of e.g. 1024 pixel, you would have approx. 253 values per pixel. (a pixel can represent maximum 1 value)

For 4 variables for a 30 days period, over 1 million variables would need to be read from the archive files.

If there are there more than just these 4 variables in the archive, the it would even need to read more. (the complete archive is read for the time period)

The first following archives would allow you to have e.g. 1 average, 1 maximum and 1 minimum value per 15 minutes for each variable, instead of having "90", 10 seconds values.

The second following archive would allow you to have 1 average, 1 maximum and 1 minimum per hour over the 15 minutes values. The third, ... the fourth, ... and so on.

[andibaciu]

ok mark ... i have the same problem and i follow your advice about following archive but i have some question ....

I must do some settings in plot graphics area for this following archive or zenon know automaticaly to use this following archive depending on period of time you select to view ???

and if i already have some archive data (old archive) without following archive linked to this data and after that i activate some following archives for this data what is the procedure to generate for this old data the following archives data ???

Thank you!

[markclemens]

I must do some settings in plot graphics area for this following archive or zenon know automaticaly to use this following archive depending on period of time you select to view ???

At the screen switch function to the extended trend screen you define the time period which should be displayed, but also the archive and the variables the data should be coming from. In this case it is not your basic archve, but the following archive. So you would have for example a "day" view function where you would see the data from the basic archive, and a e.g. a "week" view function where you would see the data from the 1st following archive in the trend.

and if i already have some archive data (old archive) without following archive linked to this data and after that i activate some following archives for this data what is the procedure to generate for this old data the following archives data ???

The data in the following archive is entered at the moment the basic archive is closed (saved). Already closed archives are not considered. The following archive is active only from the moment it is configured, and does not take existing basic archives in account.