SAP HANA Vora provides an in-memory processing engine which can scale up to thousands of nodes, both on premise and in cloud. Vora fits into the Hadoop Ecosystem and extends the Spark execution framework.


Concepts and Requirements:


Sap HANA VORA 1.2 consists of the two following main components:

  • SAP HANA Vora Engine:
    SAP HANA Vora instances hold data in memory and boost the performance.
  • SAP HANA Vora Spark Extension Library:
    • Provides access to SAP HANA Vora through Spark.
    • Makes available additional functionality, such as a hierarchy implementation.

/wp-content/uploads/2016/04/1_937827.png

These two components are included inside the Vora packages which are available as follows and you could choose based on your Hadoop distribution.

  • SAP HANA Vora for Ambari: VORA_AM<version>.TGZ
  • SAP HANA Vora for Cloudera: VORA_CL<version>.TGZ
  • SAP HANA Vora for MapR: VORA_MR<VERSION>.TGZ

/wp-content/uploads/2016/04/2_937860.png


To download the packages: https://launchpad.support.sap.com/#/softwarecenter/search/vora%25201.2


Vora 1.2 supports the following operation systems:

  • SUSE Linux Enterprise Server (SLES) 11 SP3
  • Red Hat Enterprise Linux (RHEL) 6.7 and 7.2

You should also follow the Installation and Administration guide for the compatibility pack installations: http://help.sap.com/hana_vora

Following table shows you the combination of operating system, cluster provisioning tool, and Hadoop distribution:

/wp-content/uploads/2016/04/3_937861.png

Remember that the minimal setup for Vora 1.2 is :

  • 4 cores
  • 8 GB of RAM
  • 20 GB of free disk space for HDFS data
  • Note: You can’t install Vora 1.2 on a single node

In order to have the Vora 1.2 running, you have to have the following Vora services installed and configured and I will walk you through their installment and configurations on the clusters.

  • SAP HANA Vora Base: Vora libraries and binaries. Installs on all hosts.
  • SAP HANA Vora Catalog: Vora distributed metadata store. Installs on one node and usually on DLOG node.
  • SAP HANA Vora Discovery Service: Manages service registrations and installs on all nodes. In server mode installs on 3 nodes(Max 7) and selects the bootstrapping host. In client mode, installs on all remaining nodes. Note: You can’t install DS server and client both on the same node.
  • SAP HANA Vora Distributed Log: Provides persistence for Vora Catalog. Install usually on the master node(5 nodes recommended).
  • SAP HANA Vora Thriftserver: Gateway compatible with Hive JDBC connector. Usually install on the jumpbox where DS, DLOG and Catalog servers are not installed.
  • SAP HANA Vora Tools: Web UI for Vora 1.2 modeler. Install on the same node as Vora Thriftserver.
  • SAP HANA Vora V2Server: Vora Engine. Installs on all worker nodes ( Datanodes)

 

The installation and configuration should either happen at the same time for all the services or you should follow the following order to make sure of handling the dependencies:

/wp-content/uploads/2016/04/6_937845.png

The following schema shows you the architecture for clusters with 4 nodes and the assignment of different Vora 1.2 services which we will set up in this document:

One Master node, One Server node and two workers.


Screen Shot 2016-04-07 at 1.39.42 PM.png

*** Our assumption is that you have your Hadoop clusters set up with HDFS 2.6.x or 2.7.1, ZooKeeper 3.4.6, Spark 1.5.2, Yarn cluster manager 2.7.1 components.


Installing Vora 1.2 Services:


Step 1) Adding Vora Base: You have to add Vora base on all nodes and they have to be installed as clients as shown below.

Screen Shot 2016-03-29 at 4.53.49 PM.png

Screen Shot 2016-03-29 at 4.58.24 PM.png

— no extra configuration is needed.

— you can click on the proceed button as is shown below even if you get the error since you’re not using MapReduce jobs:

Screen Shot 2016-03-29 at 4.59.49 PM.png

— Click on complete.

Screen Shot 2016-03-29 at 5.02.04 PM.png

— notice that the Vora base is now added to your services:

Screen Shot 2016-03-29 at 5.02.51 PM.png

Step 2) Now we add Vora discovery as 3 Vora discovery servers and one client.

Screen Shot 2016-03-29 at 5.03.40 PM.png

Adding the Vora Discovery client:

Screen Shot 2016-03-29 at 5.18.04 PM.png

— Vora discovery servers need extra configurations:

— in vora_discovery_bootstrap add the master DNS

— in vora_discovery_servers add your server DNS’s

Screen Shot 2016-03-29 at 5.20.32 PM.png

— proceed and deploy the service

notice that vora discovery service is now installed:

Screen Shot 2016-03-29 at 5.24.47 PM.png

Step 3) Now we add Vora Distributed Log service :


Screen Shot 2016-03-29 at 5.26.11 PM.png

— we install DLOG servers on the same machines where we installed our Discovery Servers.

Screen Shot 2016-03-29 at 5.29.49 PM.png

— No extra configurations are needed.

— click Next-> click Proceed anyway—>click  Complete

— Notice that vora DLOG is now added to the services:

Screen Shot 2016-03-29 at 5.31.47 PM.png

Step 4) Next step is to install Vora Catalog:

Screen Shot 2016-03-29 at 5.33.03 PM.png

— Install Catalog on your master node:


Screen Shot 2016-03-29 at 5.35.12 PM.png


— click Next->click Proceed anyway—>click Complete

— Notice that vora Catalog is added to the services:

Screen Shot 2016-03-29 at 5.36.55 PM.png

Step 5) Time to install V2Server as shown below:


Screen Shot 2016-03-29 at 5.38.10 PM.png


— extra configuration: add the Vora V2Server Worker service to worker1 and worker2 nodes and remove it from your server node.

Screen Shot 2016-03-29 at 5.40.45 PM.png

— click Next->click Proceed anyway—>click Complete

— Notice that vora V2Server is now added to the services:

Screen Shot 2016-03-29 at 5.43.47 PM.png

Step 6) Time to install Vora Thriftserver and Vora Tools:

Screen Shot 2016-03-29 at 5.45.13 PM.png

Screen Shot 2016-03-29 at 5.47.40 PM.png

— you have to add more configurations to the thrift server as it’s shown below:

— add vora_thriftserver_java_home = /usr/lib/jvm/java —this value depends on where JAVA installed on your system

— add vora_thriftserver_spark_home =  /usr/hdp/2.3.4.0-3485/spark this is your Spark Home value

Screen Shot 2016-03-29 at 5.51.58 PM.png

— click Next-> click Proceed anyway—>click Complete

— Notice that vora thriftServer and Vora tools  are now added to the services:

Screen Shot 2016-03-29 at 5.53.03 PM.png

Now click on HDFS, MaprReduce2 and YARN services which are in red and restart all affected as shown below:

Screen Shot 2016-03-29 at 5.58.15 PM.png

Congratulations!! You now have Vora 1.2 services installed on your clusters.

Step 7) To validate your Vora:

— SSH to your worker1 node and run:

— source /etc/vora/vora-env.sh

— $VORA_SPARK_HOME/bin/start-spark-shell.sh


and you should now see the SQL contexts (Vora SQL Context and Spark SQL Context) bieng available.

To report this post you need to login first.

16 Comments

You must be Logged on to comment or reply to a post.

  1. Asif Maqbool

    Hi Puntis

    Thanks.

    Can I install this on Quick Start VM from Cloudera 5.5 ? I believe so i can, please confirm.

    Above you have written that

    • Note: You can’t install Vora 1.2 on a single node

    But there are lots of youtube videos of how to install VORA on Single Node and even in SAP VORA Install guide does not mention anything as such that you cannot install on single node.

    Please shed some lights on this.

    Asif

    (0) 
    1. Puntis Jifroodian-Haghighi Post author

      Hi Asif,

      the videos shown on YouTube belong to the older version of Vora where we hadn’t introduced the Vora discovery services. With Vora 1.2, you need at least 3 nodes to install Vora discovery service. You can install Vora 1.2 with Cloudera 5.5.

      please also check the documentation for other dependencies.

      Thanks,

      Puntis

      (0) 
      1. Asif Maqbool

        Puntis

        Thank you.

        Can i use Cloudera Quick Start VM (I believe QVM is single node)

        or use HDP 2.3 ( Hortonworks i believe is Multi Node) ?

        Please advise.

        Asif

        (0) 
        1. Rahul Deo Vishwakarma

          Hi Asif,

          I am afraid to say that you can not use Cloudera Quick Start VM or Hortonworks Sandbox for setting up Vora. These VMs are based on CentOS.

          As you can see in the version support matrix above by Puntis, it clearly shows that CentOS is not supported OS:


          Vora 1.2 supports the following operation systems:

          • SUSE Linux Enterprise Server (SLES) 11 SP3
          • Red Hat Enterprise Linux (RHEL) 6.7 and 7.2

          You have to install Ambari / Cloudera /MapR on supported OS (RHEL or SLES11SP3). The steps for Ambari/Cloudera installation are still the same as mentioned at https://github.com/saphanaacademy/Vora/blob/master/Vora_SingleNodeInstallNotes.txt

          Before installing please check the version support matrix or the installation won’t succeed.

          Thanks,

          Rahul Vishwakarma

          (0) 
          1. Asif Maqbool

            Rahul

            Thanks for the advise.

            1) Vora 1.2 can be installed on Single Node ?

            2) The link that you have given above applies to which version of Vora ? 1.2. or previous ?

            + The link from github applies to having SLES 11 on AWS. I want to install Ambari on SLES 11 VM present on my machine.

            Please advise.

            Asif

            (0) 
            1. Rahul Deo Vishwakarma

              HI Asif,

              I have not tried Vora 1.2, but I am trying to install on Single Node. If I succeed I will surely tell you.
              The link is applicable to Vora 1.1. But, you can use the command to install Ambari 2.2 for SLES11 SP3 VM (make sure VM is allocated 8GB RAM and 2 Cores of Processor). If you notice carefully, then the git docuemtnation uses AWS to create SLES11SP3 Machine and in your case you are tryign to create SLES11SP3 as a VM. I mena to say the commands for installtion of Ambari, Spark and VOra will reamin same.

              Here is the way to Install Ambari on SLES11SP3 VM : https://cwiki.apache.org/confluence/display/AMBARI/Install+Ambari+2.2.0+from+Public+Repositories

              Use this repo for SLES11 SP3 : http://public-repo-1.hortonworks.com/ambari/suse11/2.x/updates/2.2.0.0/ambari.repo

              Once you are successful in installing Ambari, then you can refer the github links for installing VORA.

              I hope this helps.

              Thanks.

              Rahul Vishwakarma

              (0) 
              1. Puntis Jifroodian-Haghighi Post author

                Hi Rahul,

                Thanks for your comments.

                Vora 1.2 doesn’t have the hack to be installed on one single node and the reason is that discovery servers vote among themselves to choose the bootstrapping host and for that you need at least 3 nodes.

                We’ve hacked this internally to create instances on CAL and AWS but the feature is not yet available for shipment.

                (0) 
                1. Asif Maqbool

                  Hi Puntis, Rahul

                  Could any one of you share with me the vm or image that has vora single node installed ?

                  or any image with any vora version.

                  I just want to avoid the need of installation and get on to some development work.

                  Asif

                  (0) 
                  1. Rahul Deo Vishwakarma

                    Hi Asif,

                    Vora installed ovf file will be around 2.5 – 3 GB and its painful to upload.

                    You can try multinode installation with Vora 1.2
                    For downloading Vora 1.2 you need to have SAP Download access. You can request your organisation to get one for  you.

                    If you face any issue with installation, then you can reach us.

                    Thanks,

                    Rahul Vishwakarma

                    (0) 
                    1. Asif Maqbool

                      Rahul

                      I managed to install hdp 2.2 single node on a sles 11 sp3 vm.

                      I need vora 1.1 package + vora spark extensions that are mentioned in dropbox of link provided for singleNodeInstall for above 2.2.

                      The Dropbox links are not working hence the reason i am requesting

                      Asif

                      (0) 
  2. Bing Liu

    Thanks Puntis.  It is a great blog.

    It is very helpful for the people like me who are studying on VORA.

    Ref to SAP HANA Vora 1.2 Installation and Administration Guide (v1.0-20160331), the Ambari 2.2 can be used in the combination.




    (0) 
  3. Shireesha Annaboina

    Hi All,

    Its a really good blog on Vora.

    I have requirement to Install Vora 1.2 in my project. After some analysis and attending Hana academy videos for “Vora 1.1 Installion on single”.

    I have few questions,as to which of the below is a right approach.

         1)  Use Windows Machine and have  SLES 11 SP3 as VM?

       or 2) Use SLES 11 SP3 as a barbone machine  and start with HDP & Ambari installations?

    Also can you please suggest if Vora 1.2  with Ambari is  good to start with or Cloudera?

    As we could find good amount of installations steps and help available for Ambari, thought of going along with it.But wanted your confirmation as you have already installed it.

    (0) 
  4. Cristian Cordobina

    Hi all,

     

    Where should be then installed spark controller? I have installed spark controller 1.6.1 on the node which has the VORA catalog and Dlog and has problems in getting the VORA catalog when I browse VORA from HANA. I have 3 servers for Discovery in server mode, no client mode, 3 Dlog servers, etc … VORA is working well but connexion from HANA through spark controller is a problem.

     

    Hadoop-VORA.JPG

    Hana is throwing a SAP DBTech JDBC:[403]: internal error: Cannot get remote source objects. Could not get tables.

     

    VORA-HANA-error.JPG

    Best regards,

    Cristian

    (0) 
  5. pushkar deshpande

    Hi Puntis
    I am getting stuck while installing the vora base service, page doesnt load at all  when I  reach step 4 i.e(customize services) . whats the solution?

    (0) 

Leave a Reply