Thank you for the response. 

few questions/clarifications: 

   - [eventually, master components will need to live on multiple nodes 
   (self-hosting)?] what do you mean by self-hosting?
   - [This is more about how much work you want to do to set up and 
   administer the cluster than it is about master component requirements.] 
   This is actually about cost saying. Doesn't that make sense? tbh, I did not 
   quite got what were trying to say.
   - [Control plane resource requirements roughly scale with the number of 
   nodes and/or pods in the cluster]: this can be very helpful. would you 
   please give an example?
   

>    
> Thanks,
Mike

>
>    
>
On Wednesday, November 23, 2016 at 1:30:23 PM UTC-8, Daniel Smith wrote:
>
>
>
> On Wed, Nov 23, 2016 at 11:39 AM, Mike <[email protected]> wrote:
>
>> Our use case (big data) demands running few short-term (hours to days) 
>> kubernetes clusters which happen to be smaller (1 to 50 node with an 
>> average of 6). Isolation within this cluster is a must so running one 
>> bigger cluster is out of the question. because of this, we want to use the 
>> master node/s as minion node/s as well. This is where I would like to use 
>> your advice:
>>
>>    1. I totally understand that sizing depends on the specific use-case 
>>    but is there a rough cluster size at which masters MUST BE independent 
>>    nodes?
>>
>> There is no such size - eventually master components will need to live on 
> multiple nodes (self hosting). This is more about how much work you want to 
> do to set up and administer the cluster than it is about master component 
> requirements. I think I would recommend you think of the master components 
> as more of a logical control plane than a VM. It's true our scripts 
> provision things that way, but it's just for convenience.
>
>>
>>    1. Master do not seem to do much so my guess is that masters do not 
>>    require that much resources for smaller clusters. is this a correct 
>>    assumption?
>>
>> Control plane resource requirements roughly scale with the number of 
> nodes and/or pods in the cluster. 
>
>>
>>    1. Is there any guide on kubernetes components resource 
>>    consumption testing?
>>
>> I don't think so. It's potentially a pretty big topic. You may find some 
> assumptions in the code if you look for places where we estimate cache 
> sizes, but I don't think it'll be that helpful.
>  
>
>> Thank you in advance for sharing your insights.
>> Mike
>>
>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "Kubernetes user discussion and Q&A" 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/kubernetes-users.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Kubernetes user discussion and Q&A" 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/kubernetes-users.
For more options, visit https://groups.google.com/d/optout.
  • [kubernetes... Mike
    • Re: [k... 'Daniel Smith' via Kubernetes user discussion and Q&A
      • Re... Mike
        • ... 'David Oppenheimer' via Kubernetes user discussion and Q&A
        • ... 'Daniel Smith' via Kubernetes user discussion and Q&A

Reply via email to