How to properly scale while implementing aggregation bolt

[Idan Fridman]

Hi,
I am worried about scaling issues regarding my aggregation bolt.

My scenario is pretty simple:

I am sending message to Storm topology. I got one bolt which executing http
request to an external webservice and result is returned with
messageGroupId .

 When I get the result I want to aggregate(Counting by messageGroupId)

And every 3 mins to persist the current map(ill use Tick bolt for that
mission).

So for example my aggregation would be simple as that:

Map<String, Integer> messageMaps;

...

@Override
public void execute(Tuple tuple..)

{

String messageGroupId=tuple.getStringByField("messageGroupId)';

String messageGroupIdKey=messageMaps.get(messageGroupId);
if(messageGroupIdKey==null)
{
    messageMaps.put(messageGroupId,1);
}
else
{
    int counter=messageMaps.get(messageGroupIdKey);
    messageMaps.put(messageGroupIdKey,counter++);
}


Now I understand that in order to bring all executions in my topology to
this instance(to use this map) I gotta use Field-Grouping.

 But how i am going to scale that bolt? else I am overloading it

 if I have 800,000 messages with the same messageGroupId they all will go
to a specific bolt..

Thank you all.