How to store the intermediate result in a topology?

[姚驰 <ya...@163.com>]

Hi all,  when I used storm to do something like computing the active users count in the last 5 minutes, 
I wandered if there is any best practice about storing the intermediate result within the topology, I mean, 
some practical way instead of using something like a map in the examples. In my job, should I just put everything into caches 
like redis or memcached? If there are any ideas, please tell me.
Thanks for your help. 

回复：How to store the intermediate result in a topology?

[唐思成 <ja...@qq.com>]

A way to handle this kind of problem is to use a tick tuple, you can config a bolt to receive a tick tuple every 5 minutes.






------------------ 原始邮件 ------------------
发件人: "姚驰";<ya...@163.com>;
发送时间: 2014年10月28日(星期二) 晚上7:57
收件人: "user"<us...@storm.apache.org>; 

主题: How to store the intermediate result in a topology?



Hi all,  when I used storm to do something like computing the active users count in the last 5 minutes, 
I wandered if there is any best practice about storing the intermediate result within the topology, I mean, 
some practical way instead of using something like a map in the examples. In my job, should I just put everything into caches 
like redis or memcached? If there are any ideas, please tell me.
Thanks for your help.

Re:Re: How to store the intermediate result in a topology?

[姚驰 <ya...@163.com>]

Thanks~






At 2014-10-28 23:11:35, "Samit Sasan" <sa...@gmail.com> wrote:

Have a look at using HazelCast for a distributed map. 


-Samit

Re: How to store the intermediate result in a topology?

[Samit Sasan <sa...@gmail.com>]

Have a look at using HazelCast for a distributed map.

-Samit