Opened 8 years ago

Closed 7 years ago

#4686 closed task (wontfix)

Decide whether we want to archive consensus download times

Reported by: karsten Owned by: karsten
Priority: Medium Milestone:
Component: Metrics/CollecTor Version:
Severity: Keywords:
Cc: arma, ln5 Actual Points:
Parent ID: Points:
Reviewer: Sponsor:

Description

The consensus-health checker now memorizes consensus download times and displays the last 7 days in a table.

Internally, we're just appending raw download times to a file. If we want to archive them and make them publicly available, we should define some data format, specify/describe the format on the Data Formats page, and put up (monthly) tarballs containing the download times.

Let's revisit this task in a few months (e.g., February 2012?) when we have some data. We might even run a quick analysis on the data first to see if there are interesting trends.

Child Tickets

Attachments (2)

download-stats.png (70.5 KB) - added by karsten 8 years ago.
Consensus download times
download-stats-comparison.png (111.8 KB) - added by karsten 7 years ago.
Comparison of consensus download times from U.S.A. and from Europe

Download all attachments as: .zip

Change History (4)

Changed 8 years ago by karsten

Attachment: download-stats.png added

Consensus download times

comment:1 Changed 8 years ago by karsten

There, I attached a graph with consensus download times by directory authority from the past few weeks.

Does that mean we should archive these data and make them publicly available? If so, should we set up a second consensus-health checker in Europe?

Changed 7 years ago by karsten

Comparison of consensus download times from U.S.A. and from Europe

comment:2 Changed 7 years ago by karsten

Cc: ln5 added
Resolution: wontfix
Status: newclosed

Linus asked me last week if the consensus download statistics are broken, or why maatuska's min value for downloading a consensus is over 1 second. I set up a second consensus-health checker on my VM that's located in Germany and let it download the consensus from all directory authorities. See the attached graph.

These results make me think that our consensus download times aren't as meaningful as we want them to be. Even comparisons over time are hardly useful, e.g., consensus downloads from urras that get worse over time from Europe but don't change much from the U.S.A.

Closing this ticket. Please re-open when you think that archiving consensus download times does have a value.

Note: See TracTickets for help on using tickets.