Extracting and Grahping XML data

[ANSI]

Has anyone successfully taken the XML data from the System>Status>Overview and graphed it in excel?

Importing the file into excel produces an unusable table. When looking at the XML file, what are the "t" and "v" tags? I assume that is the data for a specific time. For all the files I have if there is n entries, there is always n+1 children for each row. Why?. In all the files have from this location the first child of the row, is a number around 1465566000. What does the t tag indicate and how is it different from the v tags?

Can anyone shed some light on this.

[ruc]

Unfortunately that data is not very useful once exported. You will need to manipulate it before you can use it in excel.

I will use the CPU and memory graph XML as example however the principles are the same for all other graph data exported from the dashboard.


1. Timestamps: All timestamps are in Unix Time format (i.e. number of seconds since 1 Jan 1970)
In excel I have been able to convert these timestamps to excel date/time format by using the formula: "Cell/(60*60*24)+DATE(1970,1,1)"

2. There is a metadata section that contains info on what data is available in the XML, followed by the actual data.

3. My notes on inline below after each line, prefixed by *** (I hope the format survives)


1464970140
*** Starting timestamp

60
***Time Increments in seconds

1464977400
*** Ending timestamp

122
***Number of values recorded/available

2
*** Number of parameters measured for each timestamp


CPU
*** First Parameter is CPU

Memory
*** Second parameter is memory



*** End of metadata


14649701402.6700000000e+010.0000000000e+00
***Each row in the data section has the timestamp & the CPU and memory values at that point in time.

Other notes:

1. For throughput graphs the throughput values are in MBps (B = bytes)

2. The absolute value of the graph named 'hits' is not very useful by itself. However when this data is viewed over time it is useful in seeing trends, especially any spikes or if you are introducing a new service through PCS it may help to compare the before/after values.

Hope this helps.

[ruc]

Sorry the format did not survive in previous post so trying again.

Unfortunately that data is not very useful once exported. You will need to manipulate it slightly before you can use it in excel.

I will use the CPU and memory graph exports as example however the principles are the same for all other graph exports from the dashboard.


1. Timestamps: All timestamps are in Unix Time format (i.e. number of seconds since 1 Jan 1970)
In excel I have been able to convert these timestamps to excel date/time format by using the formula: "Cell/(60*60*24)+DATE(1970,1,1)"

2. There is metadata section that contains info on what data is available in the XML, followed by the actual data.

3. My notes on inline below after each line, prefixed by *** (I hope the format survives)

[code]
1464970140
*** Starting timestamp

60
***Time Increments in seconds

1464977400
*** Ending timestamp

122
***Number of values recorded/available

2
*** Number of parameters measured for each timestamp


CPU
*** First Parameter is CPU

Memory
*** Second parameter is memory



*** End of metadata


14649701402.6700000000e+010.0000000000e+00
***Each row in the data section has the timestamp & the CPU and memory values at that point in time.

[/code]

Other notes:
1. For throughput graphs the throughput value are in MBps (B = bytes)
2. The absolute value of the graph named 'hits' is not very useful by itself. However when this data is viewed over time it is useful in seeing trends, especially any spikes or if you are introducing a new service through PCS it may help to compare the before/after values.

Hope this helps.

[ruc]

Sorry I'll try to use keywords from the XML file this time to get around the issue of the forum tool stripping away the XML tags.

[i]start[/i]
*** Starting timestamp

[i]step[/i]
***Time Increments in seconds

[i]end[/i]
*** Ending timestamp

[i]Rows[/i]
***Number of values recorded/available

[i]columns[/i]
*** Number of parameters measured for each timestamp


[i]CPU[/i]
*** First Parameter is CPU

[i]Memory[/i]
*** Second parameter is memory


[i]/meta[/i]
*** End of metadata

[i]data[/i]
***Each row in the data section has the timestamp & the CPU and memory values at that point in time.

[zanyterp]

In addition to what Ruc posted, the difference between 't' & 'v' is time in Epoch & value in scientific notation.