We’d like to do analysis across both, so preferably, yes, we’d like to do 
group-by’s on both. Or at least I think that’s what we want – minimally, we’d 
want to chart number of users for a given AP in something like 10–30 minute 
intervals. That would require both being tags, right? I mean if users were 
stored in fields, it’d mean no two users could share the same timestamp. 
Perhaps that’s not a big issue – I’ll go inspect our logs for timestamp overlap 
– but it’s something I’d conceptually rather not worry about.



On 18 November 2016 at 17:48:33, Sean Beckett 
([email protected]<mailto:[email protected]>) wrote:

It's not possible to predict accurately how much RAM you will need, but with 4 
million series (roughly a day of unique user and AP ids) you would need 
something on the order of 32-64GB for the system to be healthy. 
http://docs.influxdata.com/influxdb/v1.1/guides/hardware_sizing/

 With 30k users and 7.5k APs, the theoretical max series is about 225 million, 
which is current infeasible, although much work is being done to allow for 
cardinalities up to 1 billion. We expect the 1.2 release to have the 
experimental version of the new index.

You can also shard the series across data nodes in our commercial clustering 
offering. With a 4 node cluster, doubly replicated, each node only handles half 
the series cardinality. With a 10 node cluster, doubly replicated, each node 
only handles 1/5th of the series cardinality.

Do you need to GROUP BY both users and APs? Perhaps there's a way to structure 
your queries so that one or the other can be a field and not a tag.

On Fri, Nov 18, 2016 at 8:02 AM, 
<[email protected]<mailto:[email protected]>> wrote:
Op dinsdag 15 december 2015 02:16:02 UTC+1 schreef Sean Beckett:
> If you have highly dynamic data, that should be stored as a field, not a tag. 
> That makes it impossible to use with GROUP BY clauses, however, which I 
> suspect is a non-starter for you.

Is there a way to guesstimate the memory requirements from the cardinality? 
We're looking to analyse how people flow through our building by analysing our 
radius logs. From these, we get a measurement approx every 30 seconds per user 
that couples a user (or device) to an access point. Picking a random day from 
the logs puts us at 7.5k access points to which some 500 users connected. 
That'd get us a cardinality of about 3.75M... that's not going to end well I 
take it then. If we broaden the sample, we're likely to approach 30k users over 
the lifetime of the analysis.

--
Remember to include the version number!
---
You received this message because you are subscribed to the Google Groups 
"InfluxData" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to 
[email protected]<mailto:influxdb%[email protected]>.
To post to this group, send email to 
[email protected]<mailto:[email protected]>.
Visit this group at https://groups.google.com/group/influxdb.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/influxdb/3d709e55-7a67-44c8-9b35-79693e26e65c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.



--
Sean Beckett
Director of Support and Professional Services
InfluxDB
--
Remember to include the version number!
---
You received this message because you are subscribed to a topic in the Google 
Groups "InfluxData" group.
To unsubscribe from this topic, visit 
https://groups.google.com/d/topic/influxdb/iyCFGavNaeU/unsubscribe.
To unsubscribe from this group and all its topics, send an email to 
[email protected]<mailto:[email protected]>.
To post to this group, send email to 
[email protected]<mailto:[email protected]>.
Visit this group at https://groups.google.com/group/influxdb.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/influxdb/CALGqCvNdJTuPeuYGOQwOKpAhrFNcMcbhEMbvEGozx4EyKsswhQ%40mail.gmail.com<https://groups.google.com/d/msgid/influxdb/CALGqCvNdJTuPeuYGOQwOKpAhrFNcMcbhEMbvEGozx4EyKsswhQ%40mail.gmail.com?utm_medium=email&utm_source=footer>.
For more options, visit https://groups.google.com/d/optout.
--
Groet,
Emiliano Heyns

-- 
Remember to include the version number!
--- 
You received this message because you are subscribed to the Google Groups 
"InfluxData" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/influxdb.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/influxdb/etPan.582f7931.72099a8b.4ef%40han.nl.
For more options, visit https://groups.google.com/d/optout.

Reply via email to