Helo! But if one will use collocated processing, such as cache.invoke(), then there's no need to send large objects other (but you need to serialize and send over your closure).
Regards, -- Ilya Kasnacheev сб, 30 мар. 2019 г. в 10:11, Denis Magda <[email protected]>: > A downside of big and growing values is that they have to be deserialized, > updated, serialized all the times on an update. Plus, you will send them > over the network to your app. Finally, no way to index the lists. > > - > Denis > > > On Fri, Mar 29, 2019 at 7:30 AM wt <[email protected]> wrote: > >> Hi SCott >> >> access will be driven by data loads on the other end. I am looking at odbc >> and rest access which will be used to load data into SQL server >> >> here is a base of how it will look >> >> class data >> { >> timespan loaddate >> string item >> string value >> } >> >> class alldata >> { >> list<data> records >> } >> >> >> then when we query it will be something like this >> >> select item, value,loaddate from cache where loaddate > somevalue and key >> = >> somevalue >> >> They will likely run ETL with joins to the cache on key and date. >> >> >> >> -- >> Sent from: http://apache-ignite-users.70518.x6.nabble.com/ >> >
