If you create a cache, either in code or XML, using the minimal list of 
parameter it won’t be accessible using SQL.

There are a number of ways you can define what’s visible using SQL. You can use 
a POJO with the @QuerySqlField annotation (and the indexTypes property in the 
XML file) or define QueryEntities. See the documentarian: 
https://www.gridgain.com/docs/latest/developers-guide/SQL/indexes 

Whether you do it on the client or server side is a bit of a religious debate, 
but either works. The important thing is that the first definition to hit the 
cluster is the one that takes effect. 

The most common pattern I see with JDBC is the caches are defined server side, 
and clients connect using the thin-client driver. Thin clients just need a 
hostname and port.

However, there is also a thick-client JDBC driver. The XML here is no different 
from any other node.

Regards,
Stephen

> On 16 Jan 2020, at 12:54, narges saleh <[email protected]> wrote:
> 
> Thanks Ilya, Steve.
> 1) What do you mean by SQL enabled? Do I still need to define the POJO 
> classes for the objects/tables?
> 2) Can I specify the caches including the table definitions entirely in XML 
> config file and pass the config file to the JDBC connection? If yes, I'd 
> greatly appreciate it if you provide some small samples. Please keep in mind 
> that we have native persistence in place not a third party database.
> 
> 
> 
> On Wed, Jan 15, 2020 at 7:29 AM Ilya Kasnacheev <[email protected] 
> <mailto:[email protected]>> wrote:
> Hello!
> 
> 4) I actually think that if you specify caches in thick client's config file, 
> and they are absent on server, they will be created.
> 
> (However, they will not be changed if configuration differs)
> 
> Regards,
> -- 
> Ilya Kasnacheev
> 
> 
> ср, 15 янв. 2020 г. в 15:59, narges saleh <[email protected] 
> <mailto:[email protected]>>:
> Hi All,
> 
> I am trying to use ignite's cache grid with native persistence and prefer to 
> use JDBC for cache/db connectivity.
> 
> 1) Is this possible, in either client or server mode?
> 2) If yes, I assume, I'd need one JDBC connection per cache, as I see it is 
> possible to specify only one cache per JDBC connection. Is this right?
> 3) Is this also true if I need to join multiple tables/caches?
> 4) Can I specify my caches in XML config file and just pass the config file 
> to the JDBC connection?
> 5) Will I get the same load performance if I JDBC with streaming set to true 
> as I'd using the streamer module directly (I see that I can specify most of 
> the streamer config options on JDBC connection configuration)?
> 
> thanks.


Reply via email to