tajo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "snv" <smallnetvisi...@foxmail.com>
Subject 回复: HA of Tajo
Date Tue, 04 Mar 2014 07:08:10 GMT
3ks for  your answer 


------------------
---------------------------------------------
BestWishes!
小网客
Blog:http://snv.iteye.com/
Email:1134687142@qq.com


 




------------------ 原始邮件 ------------------
发件人: "Hyunsik Choi";<hyunsik@apache.org>;
发送时间: 2014年3月4日(星期二) 下午3:05
收件人: "dev"<dev@tajo.incubator.apache.org>; 

主题: Re: HA of Tajo



Hi 小网客,

Thank you for your interests. We have a plan to support TajoMaster HA by Q2 this year. The
reason why we firstly support TajoMaster HA is that TajoMaster only has the single point of
failures (SPOF) in the Tajo architecture. QueryMaster is launched for each query. It's influence
on the system may be trivial.

If you intended mid-query fault tolerance, currently Tajo restarts failed tasks during query
processing. This approach is similar to that of MapReduce. But, its is incomplete in some
cases. We have a plan to finish that by April.

If you give more detailed questions, I can answer in more detail.

Regards,
Hyunsik Choi

On Mar 4, 2014, at 3:20 PM, 小网客 <smallnetvisitor@foxmail.com> wrote:

> hi all
> What do you think about HA of Tajo?
> 
> 
> ------------------
> ---------------------------------------------
> BestWishes!
> Blog:http://snv.iteye.com/
> Email:1134687142@qq.com
Mime
  • Unnamed multipart/alternative (inline, 8-Bit, 0 bytes)
View raw message