Why only one trace_span data in Telecom dataset?

[yiyualt]

Hi, thank you for the excellent work on processing the original AIOps 2020 and AIOps 2021 Challenge datasets.

I had a quick question regarding the AIOps 2020 (Telecom) data. In the original version, there are six types of trace data available for each day, as shown here:

,

However, in your processed version, it seems only the trace_jdbc data has been retained. Could you share the reasoning behind this choice? I’m trying to understand how the partial trace data (e.g., cmdb_id such as docker_006 and dsName such as db_003) are meant to be related to the OS components that appear in the record.csv file as ground truth.

THANK YOU,
best,
Yi

[Siyuexi]

Hi Yi,

Thanks for your message. All the original CSV files have been merged into a single file in the processed version. If you scroll further down, you should see not only the JDBC traces but also other types such as CSF, LOCAL, and others.

Junjielong

[sun-binwen]

我遇到了同样的问题，telecom这个数据集缺失了logs目录

例如这样的报错

FileNotFoundError Traceback (most recent call last)
File :5
3 # List the files in the logs directory for April 11, 2020
4 logs_directory = 'dataset/Telecom/telemetry/2020_04_11/logs/'
----> 5 log_files = os.listdir(logs_directory)
7 log_files

[Siyuexi]

@sun-binwen
Telecom does not include log data. This subset only provides traces and metrics. More details can be found in the paper Figure 2.

And I think it might not be the same problem as the previous one, which are questioning about why only one file for all traces.