Yes you are correct in your assumption.  Apologies for not clarifying. 
 Thanks for all the info.  I will take all of this into consideration.

Thanks again.

Jamie P.

On Tuesday, June 28, 2016 at 1:56:30 PM UTC-4, Frederic Desjarlais wrote:
>
>
> I'm assuming you're referring to Graylog Server "clustering" -- and not 
> MongoDB or ElasticSearch.
>
> If "high availability" is important to you, then I'd suggest having at 
> least 2 Graylog Server processes running across as many availability zones 
> as possible (e.g. different racks in a data center, etc.)
>
> You'll also want to ensure that whatever is feeding your Graylog Servers 
> can re-establish a "connection" (a term I'll use loosely due to UDP being a 
> possible protocol) in the event of a failure with the Graylog Server it 
> initially connects with.
>
> Lastly, you should ensure that you measure the performance (e.g. CPU, 
> disk, memory, throughput, etc.) of each Graylog Server node to ensure you 
> have enough capacity.  You could also take a look at 
> https://www.graylog.org/tools/sizing-estimator for some capacity planning 
> help (but this shouldn't replace collecting performance metrics into your 
> monitoring system).
>
> In terms of 'hard limits' to the number of devices which can connect to a 
> Graylog server, it depends on which input protocol is used (e.g. TCP or 
> UDP).  For TCP-based connections, the number of file descriptors allocated 
> to the Graylog Server process will determine this limit.
>
> HTH,
> Frederic
>
>
>
>
> On Tuesday, June 28, 2016 at 10:30:19 AM UTC-7, Jamie P wrote:
>>
>> I have looked this group over and did some Google searches to no avail. 
>>  My question is, at what point do you consider using a clustered setup vs. 
>> a single server instance?  I know it's based off of how many servers and 
>> devices will be reporting to the server but I can't find any info that 
>> suggests "Well if you have this many devices and servers reporting in then 
>> you should consider having this many nodes, and this many graylog 
>> instances, etc., etc."
>>
>> I want to make sure that I build out the correct solution.  I don't want 
>> to go overkill and over estimate, but I don't want to under estimate as 
>> well.  Any documentation or websites discussing this would be most helpful. 
>>  Thanks.
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Graylog Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/graylog2/b0f56fae-5f88-498c-a3eb-bdba4bb320b9%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to