I get it. Thank you for answering my question Mikael :-) .
> 在 2018年7月30日，16:18，Mikael <mikael-arons...@telia.com> 写道：
>
> Hi!
>
> With native persistence enabled it's all done for you, a query will search
> both ram and disk contents and at a restart Ignite will load "missing" data
> from disk when needed, so if you do an SQL query it will load any data from
> disk if it should be included in the result, for the query point of view it's
> no difference if you have (native) persistence enabled or not, it works the
> same way.
>
> Mikael
>
> Den 2018-07-30 kl. 09:30, skrev Lijun Cao:
>> Hi:
>>
>> I have referenced How Ignite SQL Works document but I still have got 2
>> questions.
>>
>> The first question is, when I load data to Ignite cluster(3 nodes) in first
>> time, the data is stored both on RAM and disk(I enabled local persistence),
>> so which area does Ignite query on in this situation?
>>
>> The second question is, when I restart the cluster which I have mentioned
>> above, the data is still on disk, but not exists in the RAM, so how can I
>> load the data from disk to RAM in this situation?
>>
>> Look for your reply :-).
>
