Google Cloud Managed Service for Apache Spark Reviews & Product Details

Don't have to manage or size clusters ourselves, it just scales as needed.

Job startup is quicker than our old self-managed Spark setup too. Review collected by and hosted on G2.com.

Documentation is still a bit thin compared to other GCP services, had to dig around for some configs.

Pricing can get confusing when jobs run longer than expected, wish cost visibility was clearer upfront. Review collected by and hosted on G2.com.

1. Idle cluster deletion helps save costs when cluster is not in use.

2. Autoscaling handles peak time load efficiently.

3. GCP support team is helpful in any critical issues.

4. Java API support is great. Cluster creation, Data-ETL and cluster deletion can be done in a single pipeline. Review collected by and hosted on G2.com.

1. Cluster creation takes a few minutes of time which is not very convenient.

2. Sometimes autoscaling does not get triggered on time and pipelines fail with dataproc-agent-failure error Review collected by and hosted on G2.com.

A great tool that maybe is not as popular as AWS EMR, but that punches above its weight. An elegant implementation. Review collected by and hosted on G2.com.

Using the GCP Storage and Processing paradigm can cause troubles in getting used to on-premise Hadoop users Review collected by and hosted on G2.com.

Google cloud DataProc best suits for below:

Open Source, managed, scalable data analytics

Serverless or manage clusters on your own

Has good integrations with other GCP services

Secure /cost-effective transient cluster with per-second billing

Hadoop ecosystem in the managed GCP environment

Easy to migrate the existing on-prem Hadoop workloads along with Hive warehouses. Review collected by and hosted on G2.com.

Dislikes are below:

Scaling down has performance & other issues because of a cool-down period of the pending task.

The serverless option should be improved with SLA-based scaling

Still for streaming options go for Dataflow

it still utilizes compute engine, not GKE (Google Kubernetes engine)

Storing persistent GCS (HCFS - Hadoop compatible file system) data from Dataproc has some slowness due to the separation of compute/storage - it is our option to choose which storage for which processing

It has HDFS storage allocated with each Dataproc node cluster, but not good for huge data storage locally Review collected by and hosted on G2.com.

Dev-Ops expertise friendly, It is very easy to use. It gives almost 99.9% uptime and connection speed. A lot of documentation available on its website. A normal website user always look for cheap solutions. Google offers a 300 dollars free cloud service for 1 year which is very attractive for small website users. These are very scalable. Review collected by and hosted on G2.com.

Late entry to the Cloud marketing. Fewer features and services. Some feature are still in the Beta version. Fewer worldwide data centers. Review collected by and hosted on G2.com.

That it's serverless - no infrastructure to manage. Big win. Bonus points for having Spark (plus API's), Hive and Hadoop preinstalled on the clusters. Optionality for launching managed instance clusters is quite nice to have as well Review collected by and hosted on G2.com.

It's a tough call - either the lack of enterprise admin and monitoring capabilities, or the inability to flexibly burst beyond fixed-capacity flat-rate pricing Review collected by and hosted on G2.com.

I like the ease of use for building clusters quickly and efficiently. At the same time I can resize them at any moment in time. I have plenty of nodes so that I don't have to be concerned about pipelines outgrowing my clusters. I like how the price is based on actual use, and that they gave me a $300 credit towards my project. Review collected by and hosted on G2.com.

Sometimes it can be not user friendly and clusters can take time to re start. Otherwise I have no other complaints. Review collected by and hosted on G2.com.

The best way to manage Spark and Hadoop service that has been offered on the Google Cloud Platform. Review collected by and hosted on G2.com.

it doesn’t work with custom sources. For the standard computer, it is still the most expensive component Review collected by and hosted on G2.com.

This software is very good. Its ability to predict opportunities in sales or manufacturing from business records or data. Smooth IDE and seamless integration with multiple programming language options. Review collected by and hosted on G2.com.

Not user friendly, would be more beneficial for data scientists or someone who specializes in IT. Clusters take time to re-start when billing is re-enabled. Review collected by and hosted on G2.com.

Use of Dataproc makes the use of Hadoop Big Data solution in Cloud seamless. It brings in the capability of storing non-structured data and decoupling with Compute so that it is scalable. Review collected by and hosted on G2.com.

Key downside has been unable to create SQL frontend similar to Hive Metastore in Hadoop platform. BigQuery currently does not support AVRO and Parquet format. Review collected by and hosted on G2.com.