It’s been a bit over a week since the general availability of HDInsight Service. I’ve been kicking the tires and thought I would share some thoughts. Right off the bat I can tell you that PowerShell integration with HDInsight is going to be a huge hit! The ease of use and the responsiveness of the PowerShell environment is absolutely awesome.
What is HDInsight?
HDInsight is the 100% Apache compatible Hadoop version that runs on Microsoft technology in Windows Azure.
Why use HDInsight Service?
First and foremost, there is a deep integration between the Microsoft BI tools that your users are already used to and HDInsight Service. Second, the PowerShell extensibility makes creating, managing, and shutting down a HDInsight Service cluster so easy a caveman can do it. Third, the development experience with HDInsight means that your developers can reuse their existing .NET skill set in addition to using Java.
Microsoft BI Integration
Need to do some post map-reduce mashing up of your data? Bring it into Microsoft Excel with Power Query (ETL for the BI Masses). In two steps, you’ll be choosing the data from HDInsight that you want to bring into excel. This just works.
Here are the instructions on connecting Excel to Windows Azure HDInsight with Power Query.
After you install and configure PowerShell for HDInsight, you can manage your Windows Azure HDInsight environment from your desktop. This means that you can configure an HDInsight cluster, submit Hive and Pig queries, and extract the data to your BI environment all from the comfort of your corporate environment. This means that you can use the tools you use today to manage schedules and handle your operations. The PowerShell toolset surprised me with its ease of use. Here is an example of configuring a cluster.
Awesome feedback in PowerShell about the state of your commands:
Richer Development Experience
Want to have more control over your environment and use Visual Studio at the same time? Check out this tutorial Submit Hive Jobs using HDInsight .NET SDK. Below is a snippet of what I have going on in my VS environment with a MapReduce Job being submitted. I’ll do some additional posts about some of the pros and cons of the .NET development experience soon.
Looking for guidance around migration from SQL Server to PDW? Microsoft has provided a new migration white paper for your guidance.
In this migration guide you will learn the differences between the SQL Server and Parallel Data Warehouse database platforms, and the steps necessary to convert a SQL Server database to Parallel Data Warehouse.
Channel 9 has an amazing abundance of informative material around SQL Server, Microsoft BI, and Big Data. Saptak Sen (Microsoft) and Bill Ramos (Advaiya) have produced the newest offering of videos that cover Microsoft’s Azure and HDInsight offerings and most importantly how they integrate with Microsoft’s Business Intelligence stack. I’ve watched three of the videos so far and really enjoyed the #4 Mahout video.
Check out the entire Big Data Analytics course here:
If you want to go directly to any of the videos, here you go:
In this module, you will learn how to use Microsoft Excel Power Query with PowerPivot to mash up data from a variety of sources including Hive tables, Windows Azure Data Marketplace, and web sources. [01:45] – Power Query Excel Add-In [04:21] – Excel Power Pivot Add-In [06:01] – Demo Big Data
This module explains how to use Microsoft Excel Power View and Power Map add-ins to visualize data mash-ups from a PowerPivot model to create charts and map-based analysis. [01:36] – Excel Power View Add-In [07:31] – Demo Creating Power View Reports [12:38] – Power Map Excel Add-In [13:52] – Demo…
In this module, you will find out how to use SQOOP to perform high-speed data transfers from a Hive table on an HDInsight cluster to a Windows Azure SQL database. You will then see how create and deploy reports on Windows Azure Reporting Services. [01:11] – Working with SQOOP in Microsoft HDInsight…
This module shows how to use the Microsoft Excel Data Mining add-in along with SQL Server Analysis Services to perform key influencers and categorization data mining techniques. You’ll learn how to install and use Apache Mahout on HDInsight. [01:02] – Data Mining [07:00] – Demo Excel Data Mining…
In this module, you will learn how to use Windows Azure tables and MongoDB as NoSQL technologies for your Big Data solutions. You’ll see how to create a .Net application for accessing Azure tables. You’ll also learn how to install and use MongoDB on a server. [00:54] – Windows Azure Table Storage…
Hortonworks gets credit for the release of the month with their Hortonworks Data Platform 2.0. What I like about the HDP vernacular is that it truly is a data platform as they don’t release anything piecemeal but work tirelessly to ensure that all the parts play nicely together for you. This release of Hadoop 2.0 includes all the YARN additions along with many improvements to technologies you may have been using already like Hive and HBase.
Want to know about YARN?
Hortonworks refers to YARN as the new OS for Hadoop. It provides the flexibility for additional data processing initiatives beyond mapreduce. Additionally, YARN provides for improved management and monitoring, multi-tenancy, improved security, high availability, and improved disaster recovery.
Take a look at this link to get more information on HDP 2.0 and YARN.
What about Hive?
Hive 0.12 is included in HDP 2.0 that provides a host of improvements to Hive. Specifically, query speed and SQL compatibility were major focuses of the Hive improvements in HDP 2.0. Query plan generation, Group BY’s, and Optimizations to COUNT stick out for me. SQL support improvements include VARCHAR support, DATE support, and Truncation support. There are literally dozens of other improvements. You can get a deeper read on the improvements here.
Apache Ambari and HBase saw significant improvements. Ambari allows you to provision, manage, and monitor a cluster running on Hadoop 2, including support for NameNode High Availability. More Ambari 1.4.1 information can be found here. HBase improvements include Snapshots, support for Windows(!), and reduced mean time to recovery. Check out this page for more HBase 0.96 information.
Want to learn more?
Join Hortonworks on November 12 for a webinar outlining the YARN based architecture of HDP 2.0. They’ll discuss all the latest improvements to HDP 2.0 for technologies like Hive, Ambari, and HBase. Jump here to register for the webinar.
Obviously much has gone into Hortonworks HDP 2.0. This is a release of momentous occasion. Its impressive that they have been able to package it all up together like this in one release. They should soon have a sandbox of the GA release here. I heard it should be available sometime the week of Oct. 28th, so if you don’t see it yet keep checking back.
Looking for a solution that helps drive operational data from your SQL Server, Oracle, or DB2 Environment? Need it to be a agentless environment for both source and destination? Read on about Attunity’s new Click-2-Load solution.
Warning: I have not yet used this product so I can’t vouch for it yet. I’m simply relating what I’ve read from their materials. Once I’ve personally tested the product, I will write a review.
Essentially Attunity is providing a product that through a point and click environment you can set Attunity to do one of two things.
1. Bulk Read changes directly out of the source database and bulk load them into PDW.
2. Through the use of Attunity’s CDC product, read the Transaction Log from the source for streaming changes into PDW.
This is all done through SQL Server PDW’s Bulk Loader API which has been published for a while but this is the first vendor product that I’ve seen to take advantage of it. If anyone has used it, please comment on your experiences.
Here is the link: