Data Processing Capabilities and Limitations

Last updated 07^th December, 2022

Objective

In this document, we list all the features currently supported by the Data Processing platform. We will also give you a peek at some of our upcoming features.

Capabilities and known limits

Processing engine

At this moment, OVHcloud Data Processing supports the following engines:

• Spark 3.3.1 (Scala 2.13.7 and Hadoop 3.3.1)
• Spark 3.3.0 (Scala 2.13.7 and Hadoop 3.3.1)
• Spark 3.2.3 (Scala 2.13.7 and Hadoop 3.3.1)
• Spark 3.2.2 (Scala 2.13.7 and Hadoop 3.3.1)
• Spark 3.2.1 (Scala 2.13.7 and Hadoop 3.3.1)
• Spark 3.1.3 (Scala 2.12.4 and Hadoop 3.2.0)
• Spark 3.0.3 (Scala 2.12.4 and Hadoop 3.2.0)
• Spark 3.0.1 (Scala 2.12.4 and Hadoop 3.2.0)
• Spark 2.4.3 (Scala 2.12.4 and Hadoop 2.8.5)

It is not possible to interact with Apache Spark cluster via command line.

The Apache Spark job in Data Processing is limited to:

• Minimum 1 and maximum 60 GB of RAM per Executor/Driver (including the memory overhead)
• Minimum 1 and maximum 16 Cores of CPU per Executor/Driver
• Minimum 8 GiB and maximum 143 GiB of local storage per Executor/Driver. This resource can't be directly configured though, for each node, you will have 9 GiB of local storage per core, minus 1 GiB used by system. (eg. If your executors have 4 cores each, they will have 35 GiB of local storage each)
• Minimum 1 and maximum 10 Executor nodes

Supported languages for Apache Spark code

• Java 8
• Scala 2.12
• Python v2.7+ and v3.4+

Available ports to public network

• FTP (21)
• HTTP (443, 80, 8443, 8080, 9090)
• Kerberos (88)
• Apache HDFS (8020, 9000)
• Apache Hive (10000, 9083, 10500, 10001, 10501, 50111, 15551, 15004)
• Apache HBase (16000, 8765)
• Apache Solr (8886)
• Apache Pulsar (6650, 6651)
• Apache Kafka (9092, 9093, 6667)
• RabbitMQ (5672)
• MySQL (3306)
• PostgreSQL (5432)
• Microsoft SQL Server (1433)
• MongoDB (27017, 27018, 27019)
• ElasticSearch (9200)

OVHcloud vRack

• Data Processing can not use or access OVHcloud vRack.

Available regions

• GRA (Gravelines in France)

IP range to allow

If you need to allow a list of IPs in your application network configuration (for example to allow connections to one of your databases), know that the ODP jobs will run on hosts with IP within a range depending on the region:

• 164.132.38.64/26 for GRA (164.132.38.64 - 164.132.38.127)

Storage

• Before submitting a job, you will need to upload your job into your OVHcloud Object Storage account.
• To process data, your data can be stored in OVHcloud Object Storage or any other cloud storage or connected storage that is accessible through public internet.

Resources quotas

We distinguish three types of resources: CPU cores, RAM and Instances.

You have a limited amount of those resources that you can use simultaneously. If you have a job using all of your quotas, you will not be able to submit other jobs until it is completed. The maximum amount of each resource you can use simultaneously is fetched from your cloud project quotas. Only the maximum quotas are shared between Data Processing and your cloud project. If you run out of resource on your cloud project, it will not impact your ability to submit a job on Data Processing.

If you would like to increase the quotas for Data Processing you will need to increase your cloud project quotas. Please visit Increasing Public Cloud quota for more details on the subject.

Planned features

• Ability to set a TTL (Time To Live) on your jobs
• Availability in different OVHcloud regions

Feedback

Please send us your questions, feedback and suggestions to improve the service:

• On OVHcloud community of users on https://community.ovh.com/en/
• On our Discord in the channel #dataprocessing-spark

---

---

These guides might also interest you...