What maps to what: migrate from Hazelcast to KiviDB

What maps to what. Migrate to KiviDB: Move from Redis, Valkey, ElastiCache, Redis Cloud, Aiven, Aerospike, Hazelcast and more: getting the data out, loading i

Hazelcast	KiviDB	Notes
`IMap<K, V>` of objects	One hash per entry (`HSET customer:42 name Ana age 31`)	Or one JSON document per entry (`JSON.SET`) for nested objects.
`IMap<K, V>` of strings or numbers	One string per entry (`SET stock:7 21`)
`IMap` TTL (`put(k, v, ttl, unit)`, `time-to-live-seconds`)	`EXPIRE` / `PEXPIREAT` on the key	Per key. A hash field can have its own TTL too (`HEXPIRE`).
`IMap` max-idle (`max-idle-seconds`)	No direct equivalent	Set a TTL and refresh it on read with `GETEX key EX <seconds>` (strings) or `EXPIRE` (hashes).
`IMap` eviction (`eviction-policy` LRU/LFU, `max-size-policy`)	`maxmemory` with `maxmemory-policy allkeys-lru` or `allkeys-lfu`	One limit for the whole server, not per map.
`IMap` entry listener (added, updated, removed, evicted, expired)	Keyspace notifications (`notify-keyspace-events`) over pub/sub	At-most-once: a subscriber that is disconnected misses events. Self-hosted only for now: KiviDB Cloud cannot turn them on yet.
`IMap` entry processor	`MULTI`/`EXEC`, `WATCH`, or a Lua script in a `lua` build	Runs on the server, but there is no Java code on the server.
`IMap` predicates, `SqlPredicate`, SQL `SELECT`	Search: `FT.CREATE` an index over the hashes or JSON, then `FT.SEARCH` / `FT.AGGREGATE`	Built in. No joins.
`IMap` indexes (`addIndex`, HASH or SORTED)	Fields in an `FT.CREATE` schema (`TAG`, `NUMERIC ... SORTABLE`, `TEXT`)
Near cache	Client-side caching with `CLIENT TRACKING`	See Near cache: check your client.
`MapStore` / `MapLoader` (read-through, write-through)	No equivalent	Do it in the application: read the database on a miss, write both.
`MultiMap`	One set per key (`SADD`)	Or a list (`RPUSH`) if the MultiMap keeps duplicates (`value-collection-type: LIST`).
`ReplicatedMap`	One hash (`HSET config key value`)	Every read goes to the server; there is no copy in each member.
`IQueue`	List: `RPUSH` to offer, `LPOP` / `BLPOP` to poll or take	`BLMOVE` to a processing list for at-least-once hand-off.
`ISet`	Set (`SADD`, `SISMEMBER`)
`IList`	List (`RPUSH`, `LRANGE`, `LINDEX`)
`Ringbuffer`	Stream (`XADD ... MAXLEN ~ n`, `XRANGE`, `XREAD`)	Readers keep their own position, as with a ringbuffer sequence.
`ITopic`	Pub/sub (`PUBLISH`, `SUBSCRIBE`)	Fire and forget, as `ITopic` is.
`ReliableTopic`	Stream, with consumer groups (`XREADGROUP`, `XACK`) for at-least-once
`FencedLock` (CP subsystem)	`SET lock:<name> <token> NX PX <ms>`, released with `DELEX ... IFEQ <token>`	Not the same guarantees: see Locks.
`IAtomicLong` (CP subsystem)	`INCR`, `INCRBY`, `DECR`, `SET ... IFEQ` for compare-and-set
`PNCounter`	`INCR` / `INCRBY`
`FlakeIdGenerator`	`INCR` on a sequence key, or keep generating IDs in the application
`CardinalityEstimator`	HyperLogLog (`PFADD`, `PFCOUNT`)	Estimator state can't be copied; start counting again.
JCache (`javax.cache`)	A Redis client's cache API, or Spring Cache over Redis
Compact, Portable, `IdentifiedDataSerializable`, Java `Serializable`	Your choice of encoding: hash fields, JSON, or bytes	See Serialization.
Jet pipelines, `IExecutorService`, `IScheduledExecutorService`, `ISemaphore`, `ICountDownLatch`	No equivalent	Run that compute in your own services.
Cluster membership, partition backups	KiviDB replicas, cluster mode, or KiviDB Cloud's Pro tier