Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

On Tue, Oct 11, 2016 at 2:44 PM, Lakmal Warusawithana wrote: > Further thinking on implementation for k8s, we need to improve in 3 places. > > 1.) Need to introduce min=0 for autoscaling policies > kubectl autoscale rc foo --min=0 --max=5 --inflight-request-count=80 > > 2.)

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

On Tue, Oct 11, 2016 at 2:58 PM, Lakmal Warusawithana wrote: > > > On Tue, Oct 11, 2016 at 2:51 PM, Manjula Rathnayake > wrote: > >> Hi Lakmal, >> >> On Tue, Oct 11, 2016 at 2:44 PM, Lakmal Warusawithana >> wrote: >> >>> Further thinking on

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

On Tue, Oct 11, 2016 at 2:51 PM, Manjula Rathnayake wrote: > Hi Lakmal, > > On Tue, Oct 11, 2016 at 2:44 PM, Lakmal Warusawithana > wrote: > >> Further thinking on implementation for k8s, we need to improve in 3 >> places. >> >> 1.) Need to introduce min=0

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

Hi Lakmal, On Tue, Oct 11, 2016 at 2:44 PM, Lakmal Warusawithana wrote: > Further thinking on implementation for k8s, we need to improve in 3 places. > > 1.) Need to introduce min=0 for autoscaling policies > kubectl autoscale rc foo --min=0 --max=5

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

Further thinking on implementation for k8s, we need to improve in 3 places. 1.) Need to introduce min=0 for autoscaling policies kubectl autoscale rc foo --min=0 --max=5 --inflight-request-count=80 2.) Have to config auto scaler for use load balancing factor (inflight-request-count) - K8S

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

If we think current solution we proposed for container based deployment without this hot pool concept still we may need some intelligence at load balancer level. Isn't it? Let say i send request to gateway.sanjeewa.info.wso2.com. Then load balancer should this request comes to gateway. Then

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

On Mon, Oct 10, 2016 at 8:01 PM, Sanjeewa Malalgoda wrote: > > When we do container based deployment standard approach we discussed so > far was, > >- At the first request check the tenant and service from URL and do >lookup for running instances. >- If matching

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

Basically this has 3 parts. 1. Enable network load balancer to dynamically create containers based on first request 2. Terminate containers if idle. 3. Improve startup time by having pool of running containers and dynamically allocate to tenants ( this is what Sanjeewa mention)

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

On Tue, Oct 11, 2016 at 10:22 AM, Lakmal Warusawithana wrote: > > On Tue, Oct 11, 2016 at 10:06 AM, Manjula Rathnayake > wrote: > >> Hi Sanjeewa, >> >> Are you suggesting an API manager deployment pattern using containers? >> Container per tenant and per

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

On Tue, Oct 11, 2016 at 10:06 AM, Manjula Rathnayake wrote: > Hi Sanjeewa, > > Are you suggesting an API manager deployment pattern using containers? > Container per tenant and per gateway, key manager etc? > Yes. With C5 APIM we will have per tenant gateway, store,

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

Hi Sanjeewa, Are you suggesting an API manager deployment pattern using containers? Container per tenant and per gateway, key manager etc? thank you. On Mon, Oct 10, 2016 at 9:06 PM, Malaka Silva wrote: > Hi Sanjeewa, > > My understanding is gateway pool is not tenant

Re: [Architecture] Speedup traffic serving process in scalable/ containerized deployment

Hi Sanjeewa, My understanding is gateway pool is not tenant specific and will not be returned but rather terminated? On Mon, Oct 10, 2016 at 8:01 PM, Sanjeewa Malalgoda wrote: > Hi All, > Starting this mail thread to continue discussion on "speedup instance > activate time

[Architecture] Speedup traffic serving process in scalable/ containerized deployment

Hi All, Starting this mail thread to continue discussion on "speedup instance activate time when we move ahead with container based deployments". As of now all of us are working on speedup server start time and deploy instances on demand with the help of load balancer. Please note that this is not