You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hadoop.apache.org by Anfernee Xu <an...@gmail.com> on 2016/01/26 08:03:43 UTC
Provisioning a physical host into an existing HDFS/Yarn cluster
Hi,
I recently got a powerful physical host, usually I get used to provisioning
the host with VM's and add them to my existing HDFS/Yarn cluster(consists
of 300+ VM's), now I'm exploring Docker based approach, so I want to know
if there're some best practices I can follow down the path.
--
--Anfernee
Re: Provisioning a physical host into an existing HDFS/Yarn cluster
Posted by Namikaze Minato <ll...@gmail.com>.
Well, since worker nodes are usually using all of their RAM and CPU,
plus generating an awful lot of I/O, you ***DEFINITELY*** shouldn't do
a cluster based on VMs.
Or you could go all the way making sure that all VMs could handle the
100% load but that would be pointless work that would amount to the
same processing power as the physical hosts without virtual machines.
Or maybe I'm missing a critical point here?
Regards,
LLoyd
On 26 January 2016 at 08:03, Anfernee Xu <an...@gmail.com> wrote:
>
> Hi,
>
> I recently got a powerful physical host, usually I get used to provisioning the host with VM's and add them to my existing HDFS/Yarn cluster(consists of 300+ VM's), now I'm exploring Docker based approach, so I want to know if there're some best practices I can follow down the path.
>
>
>
> --
> --Anfernee
---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@hadoop.apache.org
For additional commands, e-mail: user-help@hadoop.apache.org
Re: Provisioning a physical host into an existing HDFS/Yarn cluster
Posted by Namikaze Minato <ll...@gmail.com>.
Well, since worker nodes are usually using all of their RAM and CPU,
plus generating an awful lot of I/O, you ***DEFINITELY*** shouldn't do
a cluster based on VMs.
Or you could go all the way making sure that all VMs could handle the
100% load but that would be pointless work that would amount to the
same processing power as the physical hosts without virtual machines.
Or maybe I'm missing a critical point here?
Regards,
LLoyd
On 26 January 2016 at 08:03, Anfernee Xu <an...@gmail.com> wrote:
>
> Hi,
>
> I recently got a powerful physical host, usually I get used to provisioning the host with VM's and add them to my existing HDFS/Yarn cluster(consists of 300+ VM's), now I'm exploring Docker based approach, so I want to know if there're some best practices I can follow down the path.
>
>
>
> --
> --Anfernee
---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@hadoop.apache.org
For additional commands, e-mail: user-help@hadoop.apache.org
Re: Provisioning a physical host into an existing HDFS/Yarn cluster
Posted by Namikaze Minato <ll...@gmail.com>.
Well, since worker nodes are usually using all of their RAM and CPU,
plus generating an awful lot of I/O, you ***DEFINITELY*** shouldn't do
a cluster based on VMs.
Or you could go all the way making sure that all VMs could handle the
100% load but that would be pointless work that would amount to the
same processing power as the physical hosts without virtual machines.
Or maybe I'm missing a critical point here?
Regards,
LLoyd
On 26 January 2016 at 08:03, Anfernee Xu <an...@gmail.com> wrote:
>
> Hi,
>
> I recently got a powerful physical host, usually I get used to provisioning the host with VM's and add them to my existing HDFS/Yarn cluster(consists of 300+ VM's), now I'm exploring Docker based approach, so I want to know if there're some best practices I can follow down the path.
>
>
>
> --
> --Anfernee
---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@hadoop.apache.org
For additional commands, e-mail: user-help@hadoop.apache.org
Re: Provisioning a physical host into an existing HDFS/Yarn cluster
Posted by Namikaze Minato <ll...@gmail.com>.
Well, since worker nodes are usually using all of their RAM and CPU,
plus generating an awful lot of I/O, you ***DEFINITELY*** shouldn't do
a cluster based on VMs.
Or you could go all the way making sure that all VMs could handle the
100% load but that would be pointless work that would amount to the
same processing power as the physical hosts without virtual machines.
Or maybe I'm missing a critical point here?
Regards,
LLoyd
On 26 January 2016 at 08:03, Anfernee Xu <an...@gmail.com> wrote:
>
> Hi,
>
> I recently got a powerful physical host, usually I get used to provisioning the host with VM's and add them to my existing HDFS/Yarn cluster(consists of 300+ VM's), now I'm exploring Docker based approach, so I want to know if there're some best practices I can follow down the path.
>
>
>
> --
> --Anfernee
---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@hadoop.apache.org
For additional commands, e-mail: user-help@hadoop.apache.org