Problem: how to store and querry massive amounts of high quality sensor data ingested in real-time from many sensors.
Properties paramount for a TSMS: distribution; stream processing; compression; efficient retrieval; approximate query processing; extensibility;
Niche: we found no TSMS using multi-model compression for our survey; Also, the existing methods do not provide all the properties listed above. They either provide no latency guarantees [22, 40], require a trade-off between latency and compression [37], or limit the supported model types [22, 40]. ModelarDB, in contrast, provides all these features.
Architecture: 3 sets of components
data ingestion;
query processing
segment storage.
来源：https://www.cnblogs.com/dulun/p/11937634.html