Hi Jacek,
Yes, the assembly would necessarily be on a single czar but we could
launch as many as we need to handle the load. Using multiple czars for a
single query assembly is an interesting idea but I don't think qserv is
structured in a way that would make that easy.
Andy
On Fri, 4 Sep 2015, Jacek Becla wrote:
> Andy
>
> So you want to separate query launching from result handling.
> Yes we can think about it, we considered it at some point.
>
> But we'd still handle all the 1000's of pieces in a single place,
> not on multiple result-czars, right?
>
> Jacek
>
>
>
>
> On 09/04/2015 01:52 PM, Andrew Hanushevsky wrote:
>> Hi Jacek,
>>
>> On Fri, 4 Sep 2015, Jacek Becla wrote:
>>
>>> Andy
>>>
>>> I am not sure I see the connection between assigning query
>>> to a different czar and your comment "if we explode queries
>>> into 1000's of pieces and then try to handle reassembly in
>>> a single place"
>> As we saw, a czar can easily become overwhelmed getting back all the
>> results from a sharded query. My take is that the number of czars needed
>> to launch queries is substantially less than the number of "czars" neeed
>> to get the results. That's what I was alluding to.
>>
>> Andy
>
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the QSERV-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1
|