Top Rated Druid Alternatives
31 Druid Reviews
Overall Review Sentiment for Druid
Log in to view review sentiment.
1. Pre-aggregate capability which allows to pre-calculate aggregations and save aggregates in segments. Thus, reduces compute and storage costs.
2. Druid UI (0.14+) which has many improvements and allows creating ingestion_specs via UI
3. REST interface for druid_broker for communication, makes it easy to integrate with microservices
4. Druid is a NoSQL DB still it has SQL query support and BAs/Analysts are comfortable using SQL to query Druid
5. Data Security options - Basic HTTP Auth and LDAP supported Review collected by and hosted on G2.com.
1. Complex Architecture - Steep learning curve and has 6 core services which makes deployment & management of Druid cluster complex
2. Memory intensive Historical services - Druid services are quite memory intensive and requires high compute+memory cloud instances.
3. Indexing support - Druid supports only 1 indexing (Inverted Index) which limits the idea of optimizing datasources as per usecase Review collected by and hosted on G2.com.
Druid is amazingly fast and has built-in connectors for most of the popular datasources .
It supports variety of dashboards which makes druid a perfect choice for any Real Time Streaming Application . Review collected by and hosted on G2.com.
Druid natively queries in Json format which is hard to pick up for a SQL user.
Rollover queries are not dynamic . Example - If you want to roll up for a specific time of one day to a specific time of another day , that might not be possible .
Web GUI is also not so user friendly for a business user .
Missing operations friendly cluster manager console.
Druid needs a dedicated server and cannot utilise existing Hadoop resources. Review collected by and hosted on G2.com.
Druid is best for low latency analytics, as it combines the best qualities of a column store and inverted indexing. With column stores, the druid can minimize I/O costs for analytical queries.
It supports OLTP and OLAP.
Real-Time Aggregation.
Batch & Real-Time Ingestion Review collected by and hosted on G2.com.
1. No fault-tolerance on the query execution path. ex: A single query be processed on hundreds of historical nodes — it completely lacks any fault-tolerance on the query execution path.
2. Straggling sub-queries on the historical nodes takes a lot of time.
3. Back filling takes lot of time. But its understandable as to update old segment and update it takes lot of time. I wouldn't consider it as a drawback.
4. As Druid Brokers need to keep the view of the whole cluster in memory , it require significantly more memory and also cause lot lot JVM GC pause.
5. In case of large queries, it saturate the processing capacity of the entire historical layer for up to tens of seconds. Review collected by and hosted on G2.com.
It is easy to integrate with other database engines like MySQL. That is called integration features are good! Review collected by and hosted on G2.com.
It will be unable to configure with some of the data analytics platforms like "Metatron." Metatron uses a modified version of druid! Review collected by and hosted on G2.com.
The community behind Druid and its docs are great. The scale at which Druid can ingest and query data is impressive. Review collected by and hosted on G2.com.
Only recent versions have support for joins between data sources. Some log messages could be more verbose. Review collected by and hosted on G2.com.
It excellently supports horizontal scalability, The deep storage functionality improves data resilience and makes it easy to add a new node. Since the data is partitioned by time out of the box, time-based queries perform exceedingly well. It can ingest a large amount of data very quickly. It has multiple plugins to suffice your need and it can integrate with many cloud infrastructure out of the box. Review collected by and hosted on G2.com.
Need to provide better features to accommodate multi-tenants. Updates to existing data are currently supported by rebuilding the corresponding time segment entirely from the true source, Instead, it should support tenant id based updates. Same-day updates are a little bit tricky and need to iron it out.
One of the places we use it to calculate demographic-based suppression of data and it is slow in that particular scenario. Review collected by and hosted on G2.com.
Apache Druid works very well if you need basic aggregations across immutable time series data. It has some really useful approximations such as HyperLogLog for fast cardinality estimations that converge to exact counts for small datasets. It also now supports Druid Sql as a query language which doesn't have the steep learning curve native Druid query language requires. Review collected by and hosted on G2.com.
Apache Druid becomes hard to use and very inefficient when your data is 1) updated 2) ingested out of order (based on timestamp) or 3) requires joins. Unfortunately this greatly limits the number of use-cases that Druid readily supports. Tooling can be built around it to support things like out of order ingestion but it makes Druid very inefficient.
Druid also has inherent bottlenecks in its design: each cluster can have only one coordinator and one overlord. We found that this made it impossible to scale a single cluster out to meet our needs. Review collected by and hosted on G2.com.
Real-time ingestion and querying capability
Sub-second query performance
Time Series based datastore
Slice N Dice support
Data Compression Review collected by and hosted on G2.com.
Inability to support nested data
Partial Join Support
Setup to bring it up for the first time Review collected by and hosted on G2.com.
1) Pre-rolled up data into dimension and metrics
2) Lighting fast data/ query result retrieval Review collected by and hosted on G2.com.
Managing the broker/cluster if load is high
Limitation in dynamic dimensions Review collected by and hosted on G2.com.
