Skip to main content

Why to use Azure Databricks to run your Apache Spark Notebooks

A few months ago (almost one year) a new flavor of Apache Spark appeared on Microsoft Azure. Azure Databricks it is just a platform optimized for Azure, where Apache Spark can run.
The consumer does not need anymore to configure the Apache Spark cluster (VM creation, configuration, network, security, storage and many more). Azure Databricks already has a cluster that is configured and ready to be used. We can focus on our application and business requirements and less on the infrastructure part.

Capabilities and Features
All the features that we have inside Apache Spark can also be found inside Azure Databricks. The same version of Spark that you have on-premises runs on top of Azure Databricks, the only difference is at the infrastructure level, where you already have the system preconfigured. As a user, you are free to scale up or down your cluster in a 'drag and drop' manner, without having to pay attention or do anything else.
All the 5 main components of Spark can be found on top of Azure Databricks:

  • Streaming Capability
  • GraphX for data exploration and cognitive analytics
  • MLib for machine learning support
  • Core API for integration with R, Python, Scala, Java, and SQL
  • DataFrames with Spark SQL for working with structured data
A nice feature of Azure Databricks is the capability to remove/terminate a Spark cluster in the moment when the cluster it is not used anymore. When you create a new cluster, you have the option to specify after how many minutes of inactivity the cluster should be terminated. This option is useful especially during the developing phase where you can save a lot of resources in this way without complicated procedures. 

Integration points
At this moment in time, there are integration points with Hadoop Storage, Apache Kafka, Azure SQL Data Warehouse, Azure Data Lake Store and Azure Blob Storage. This in combination with Spark capability to combine data stream with static data makes Apache Spark a super exciting option when you need fast deployment and reliable services. 
As a side note, Azure Databricks it is using Azure Container Services behind the scene.

Scenarios
Even if Spark has an excellent support for ML, Data Analysis and Fog Computing in IoT scenarios, in the most of the cases, Spark it is used for the streaming capabilities, where it is enabling us to ETL on top of data streams, trigger events based on stream content or data enrichment with static content when needed. 
It is a clear trend to do a lot of AI and ML inside Azure Spark that enables us to process and analyze data streams in real-time. 

Pricing
The pricing model of Azure Databricks contains two components:
  1. Azure VM price
  2. DBU (Databricks Unit)
Depending on the cluster size, you will pay for the no. of Azure VM's that you are using inside the cluster. You can specify the type and the number of the VM's that you want to use. The price of the VMs it's the same as for Azure VM with Linux.
DBU represents the processing capability. Based on the type of the VM you will notice that each VM has a different DBU capability. D12 has 1 DBU vs D15 that had 5 DBUs for each node. For each DBU you will an additional amount of money. We can imagine that inside DBU are all other costs of cluster management from configuration and networking to storage and environment optimization. 
For DBU there are two pricing tiers (Standard and Premium). The most significant difference between this two is the Role Base Access control that can be found on the Premium tier. 

What you shall take a look on Azure Databricks?
  1. Autoscaling capability
  2. Auto-termination capability
  3. Optimized by design
  4. Connectors with Azure Services (e.g. Azure Data Lake and Blob Storage)
  5. PowerBI integration
  6. Support for automation
  7. Simple price model
  8. Multiple types of Azure VM supported
  9. Enterprise-ready with Azure AD support
  10. Security and privacy out of the box

Comments

Popular posts from this blog

Windows Docker Containers can make WIN32 API calls, use COM and ASP.NET WebForms

After the last post , I received two interesting questions related to Docker and Windows. People were interested if we do Win32 API calls from a Docker container and if there is support for COM. WIN32 Support To test calls to WIN32 API, let’s try to populate SYSTEM_INFO class. [StructLayout(LayoutKind.Sequential)] public struct SYSTEM_INFO { public uint dwOemId; public uint dwPageSize; public uint lpMinimumApplicationAddress; public uint lpMaximumApplicationAddress; public uint dwActiveProcessorMask; public uint dwNumberOfProcessors; public uint dwProcessorType; public uint dwAllocationGranularity; public uint dwProcessorLevel; public uint dwProcessorRevision; } ... [DllImport("kernel32")] static extern void GetSystemInfo(ref SYSTEM_INFO pSI); ... SYSTEM_INFO pSI = new SYSTEM_INFO(

Azure AD and AWS Cognito side-by-side

In the last few weeks, I was involved in multiple opportunities on Microsoft Azure and Amazon, where we had to analyse AWS Cognito, Azure AD and other solutions that are available on the market. I decided to consolidate in one post all features and differences that I identified for both of them that we should need to take into account. Take into account that Azure AD is an identity and access management services well integrated with Microsoft stack. In comparison, AWS Cognito is just a user sign-up, sign-in and access control and nothing more. The focus is not on the main features, is more on small things that can make a difference when you want to decide where we want to store and manage our users.  This information might be useful in the future when we need to decide where we want to keep and manage our users.  Feature Azure AD (B2C, B2C) AWS Cognito Access token lifetime Default 1h – the value is configurable 1h – cannot be modified

What to do when you hit the throughput limits of Azure Storage (Blobs)

In this post we will talk about how we can detect when we hit a throughput limit of Azure Storage and what we can do in that moment. Context If we take a look on Scalability Targets of Azure Storage ( https://azure.microsoft.com/en-us/documentation/articles/storage-scalability-targets/ ) we will observe that the limits are prety high. But, based on our business logic we can end up at this limits. If you create a system that is hitted by a high number of device, you can hit easily the total number of requests rate that can be done on a Storage Account. This limits on Azure is 20.000 IOPS (entities or messages per second) where (and this is very important) the size of the request is 1KB. Normally, if you make a load tests where 20.000 clients will hit different blobs storages from the same Azure Storage Account, this limits can be reached. How we can detect this problem? From client, we can detect that this limits was reached based on the HTTP error code that is returned by HTTP