Limited Uplink Usage History Despite Lookback Period

[Eddysanoli]

Hello everyone,

Im trying to create a feature that tracks uplink usage across multiple months. Currently Im using the following endpoint:

https://developer.cisco.com/meraki/api-v1/get-network-appliance-uplinks-usage-history/

However, Im encountering a limitation. The documentation mentions that the parameter "t0" has a maximum lookback period of 365 days from today, so a year. And although this is technically true (the endpoint does return timestamps for up to a year back), it returns no measurements. The endpoint simply returns no interfaces after going back a month or so. I have verified that this is the case with 10 different networks, all with multiple years of activity so far. 

Is this a known limitation of the endpoint? Or is there a detail that Im not aware of? I have made sure to also test all of the available resolutions, and the issue remains in each one of them

I would appreciate any help of clarification on this issue.

Thanks in advance

[RWelch]

Probably more along the lines of data retention limitations.  Dashboard Data Availability 

[Oren · Accepted solution]

It's a mistake in the documentation. It should say 31 days.
I'll make sure the team updates it.

[Eddysanoli]

Thank you! Do you think it would be possible to retrieve data further back at any point? Or its as @RWelch mentioned, a data availability issue?

[Oren]

I am not aware of such plans.
What is the use-case for polling data over a month old?

You could collect that information using Splunk, or one of the many monitoring ecosystem partners we have on the ecosystem marketplace.

[Eddysanoli]

Okay, no worries. We needed to analyze past months for instances of overuse, so that the property could be billed for a higher bandwidth. Unfortunately adding something like splunk would be impossible and too much of a large scale operation with our current situation.

Thank you

[sungod]

When I've set up reporting for this type of thing, the general method is to run daily or monthly scripts (automatically) to build up the data.

Then it's available locally to analyse whenever you need.

[MTaylor]

When I started using this API endpoint it returned up to 365 days of data. 
This changed around the first of the year (obviously a bug) but now its a documentation mistake. 
Here's one use case: We host annual events and need to look back at the utilization from the previous year.

[sungod]

Even if it were 365 days, which I assume would be tested as 31536000 seconds internally...

- you'd still need to cope with leap years by separately acquiring an extra day

- 365 'days' from when? 00:00:00 UTC? each local network's time zone? individual device clocks? some timestamp in a table? etc. - this sort of thing is not well defined across the API documentation (this is not unique to Meraki)

- devices do not report instantly, there's a delay

- unless you could nail acquisition at exactly 365 days since last time, you'll lose data or maybe have an overlap, due to latency, clock errors, rate limits, outages etc.

Because I'm picky about time, I spent a while digging into this stuff when creating a platform that acquires/analyses data from Meraki and other SDN platforms.

I decided it's better overall to acquire in shorter chunks month/day/whatever, then there's margin to recover from problems, and avoid edge effects due to retention period cut-offs.