Hazelcast IMap性能

Question

I use the hazelcast 3.8.6. There is a IMAP{Key, Value} with next settings:

mapConfig.setBackupCount(5);
mapConfig.setReadBackupData(true);

the Key is

long userId;
long contentId;

also there are 5 nodes where the data is. As far as i understand according the setting each node will be have all data. Such way i turn a distributed map to a replicated map to increase speed of get operations and don't do networking calls.

let's image i need get all data for user by id. in usual mode i would write next code:

EntryObject eo = new PredicateBuilder().getEntryObject();
Predicate predicate = eo.key().get("this.userId").equal(userId);
map.values(predicate).stream().map(.....)

this code will be executed on all nodes and result will be returned to node which make the call. Am i right understding that in case of the settings "setReadBackupData(true)" the request won't be send to other nodes and all data will be found on node wich make this request? or in that case to avoid networking calls i just should use something like this:

map.entrySet().stream()
.filter(entry->entry.getKey().getUserId().equals(userId))
.map(.....)

thanks.

Answer 1

请记住，仅当您从Member内部执行查询时，readBackupData标志才起作用，因此，如果您使用的是HazelcastClient，则它将不会从备份中读取。