[prev in list] [next in list] [prev in thread] [next in thread] 

List:       activemq-users
Subject:    Re: DR: Queue backup / replay architecture (Artemis)
From:       <warm-sun2 () tutanota ! com>
Date:       2019-07-08 23:33:07
Message-ID: LjIvk69--3-1 () tutanota ! com
[Download RAW message or body]


I wasn't clear enough ;)

We are a service provider.
This particular DR scenario is: if the clients lose the data/transactions (on their \
end) and need a replay of the transactions (messages). Clients have independent \
platforms to us.

(Our DR, on our sites, is having 2 data centers and replication)

On traditional brokers: messages are designed to be sent and forgotten...
In this scenario, described above, we are sending the message -- but want to keep a \
backup for 1 month, in case the client will need a replay.

Is there any other more suitable pattern for this scenario? (I can't think of any and \
didn't find any in my research, eg EIP: Enterprise Integration Patterns). Any other \
pattern would involve consuming messages from the broker and storing them in some \
other back-end and if the clients ask for a replay -- re-enqueing the messages back \
on to the broker, which would be inefficient (compared to the ones I'm proposing). \
Would be interested in your opinions.

Thanks for answering all the other questions so perfectly!

Also, do you think having an option to shallow-copy messages in divert use cases like \
the above is a valid feature for Artemis? It would save 1/2 the disk space in our use \
case, because we are not transforming message bodies.



[prev in list] [next in list] [prev in thread] [next in thread]