BlueData Isilon Validation Brief

Technical Guide – ISV Partner Validation
TECHNICAL VALIDATION OF
BLUEDATA EPIC WITH EMC ISILON
ETD Solution Architecture Compatibility and
Performance Testing Validation Brief
Boni Bruno, CISSP, CISM, CGEIT
Chief Solutions Architect
ABSTRACT
This document captures details on the technologies, results, and environment used to
perform various functionality tests to demonstrate compatibility between BlueData
EPIC Software and EMC technologies described herein.
November 3 2015
Revision 1.3

EMC Corporation 2
To learn more about how EMC products, services, and solutions can help solve your business and IT challenges, contact your local
representative or authorized reseller, visit www.emc.com, or explore and compare products in the EMC Store
Copyright © 2015 EMC Corporation. All Rights Reserved.
EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without
notice.
The information in this publication is provided “as is.” EMC Corporation makes no representations or warranties of any kind with
respect to the information in this publication, and specifically disclaims implied warranties of merchantability or fitness for a
particular purpose.
Use, copying, and distribution of any EMC software described in this publication requires an applicable software license.
For the most up-to-date listing of EMC product names, see EMC Corporation Trademarks on EMC.com.
EMC are registered trademarks or trademarks of EMC, Inc. in the United States and/or other jurisdictions. All other trademarks used
herein are the property of their respective owners.
Part Number H#pending

EMC Corporation 3
TABLE OF CONTENTS
INTRODUCTION........................................................................................5
Purpose ............................................................................................................ 5
Bluedata and Isilon joint value proposition ............................................................ 5
Technology Overview.......................................................................................... 5
Relevant Partner Features ................................................................................... 7
Relevant EMC Features ....................................................................................... 7
VALIDATION SCOPE .................................................................................7
Partner Technologies .......................................................................................... 7
EMC Technologies .............................................................................................. 7
TESTING ENVIRONMENT...........................................................................7
Architecture Illustration ...................................................................................... 7
Resources ......................................................................................................... 8
Bluedata EPIC documentation.............................................................................. 9
BLUEDATA installation and Configuration .............................................................. 9
EMC ISILON Installation and Configuration.......................................................... 15
Supporting Information and Considerations ......................................................... 21
VALIDATION TESTS ................................................................................21
Compatibility Testing ........................................................................................ 21
TEST # 1: Connect to Kerberized Isilon HDFS via BlueData DataTap object. ............ 21
TEST # 2: Connect to Isilon via BlueData NFS DataTap......................................... 24
TEST # 3: Browse Kerberized Isilon HDFS DataTap from CDH (Cloudera) Compute
Cluster............................................................................................................ 25
TEST # 4: Browse Kerberized Isilon HDFS DataTap from HDP (Hortonworks) Compute
Cluster............................................................................................................ 27
TEST # 5: Run a Hive query against data stored in Isilon Kerberized HDFS via Hue
console ........................................................................................................... 28
TEST # 5: Integrate with Cloudera Navigator for metadata search and auditing....... 31
TEST # 6: Preliminary performance benchmarks using standard development/test
hardware in BlueData Labs................................................................................ 37
Testing Engineers & Testing Acceptance.............................................................. 41

EMC Corporation 4
APPENDIX ..............................................................................................42
Teragen command ........................................................................................... 42
Terasort command ........................................................................................... 43
Teravalidate command...................................................................................... 44
TEST # 7: Hadoop Cluster (CDH) with Isilon as HDFS replacement ........................ 45
Comparison of Cloudera on BlueData + Isilon vs. Cloudera Standalone + Isilon
(Navigator) ..................................................................................................... 48
1. Metadata – Hive Query................................................................................ 48
2. Metadata – Impala Query ............................................................................ 51
3. Metadata – Pig Query.................................................................................. 52
4. Metadata – Oozie ....................................................................................... 54

EMC Corporatoin 5
INTRODUCTION
PURPOSE
The intent of this document is to capture details pertaining to the technologies and environment used to confirm compatibility
between BlueData EPIC Software and EMC Isilon storage system
BLUEDATA AND ISILON JOINT VALUE PROPOSITION
The BlueData EPIC™ (Elastic Private Instant Clusters) software platform solves the infrastructure challenges and limitations that can
slow down and stall Big Data deployments. With EPIC software, you can spin up Hadoop or Spark clusters – with the data and
analytical tools that your data scientists need – in minutes rather than months. Leveraging the power of containers and the
performance of bare-metal, EPIC delivers speed, agility, and cost-efficiency for Big Data infrastructure. It works with all of the major
Apache Hadoop distributions as well as Apache Spark. It integrates with each of the leading analytical applications, so your data
scientists can use the tools they prefer. You can run it with any shared storage environment, so you don’t have to move your data.
And it delivers the enterprise-grade security and governance that your IT teams require. With the BlueData EPIC software platform,
you can provide Hadoop-as-a-Service or Spark-as-a-Service in an on-premises deployment model.
EMC Isilon Scale-out Storage Solutions for Hadoop combine a powerful yet simple and highly efficient storage platform with native
Hadoop integration that allows you to accelerate analytics, gain new flexibility, and avoid the costs of a separate Hadoop
infrastructure. BlueData EPIC Software combined with EMC Isilon shared storage provides a comprehensive solution for compute +
storage.
BlueData and Isilon share several joint customers and opportunities at leading financial services, advanced research laboratories,
healthcare and media/communication organizations. As such, the following diagram describes the key value propositions:
This paper describes the process of validating Hadoop applications running in virtual clusters on the EPIC platform with data stored
on the EMC Isilon storage device using either NFS or HDFS data access protocols.
TECHNOLOGY OVERVIEW
The BlueData EPIC platform is an infrastructure software platform, purpose built for to virtualize/containerize Big Data/Hadoop
applications to increase agility, significantly reduce infrastructure cost by increasing utilization while maintaining security and control.
As such, BlueData software enables Big Data-as-a-service on-premises.
BlueData + Isilon è Lowest TCO + Max Agility
HDFS DAS Isilon
Physical Hadoop Compute
Infrastructure
(e.g. MR, Hive, Impala etc)
BlueData
Containerized
Infrastructure
NFSHDFSHDFS
Single Distribution & Version; No Elasticity Multiple Distribution & Versions; Elasticity
$$ compute savings
$$ storage savings
BlueData DataTap (dtap://….)
Federate across
DAS & Isilon for
frictionless
transition to Isilon
ü Bring elastic Hadoop and Spark compute to Isilon with NFS/HDFS DataTap
ü Frictionless transition from DAS to Isilon environments with single namespace
ü Future proof against rapidly evolving ecosystem – e.g. Cloudera Navigator, Ranger

EMC Corporation 6
The point of interconnection with EMC Isilon storage is the EPIC DataTap feature. DataTap surfaces an HDFS abstraction layer so that
any Big Data application can be run unmodified, but at the same time implements a high-performance connection to remote data
storage systems via NFS, HDFS, and object RESTful protocols. The result allows unmodified Hadoop applications to run against data
stored in remote NFS, HDFS, and object storage systems without loss of performance.

EMC Corporation 7
RELEVANT PARTNER FEATURES
Table 1 List of Partner technology features relevant to the validation consideration
DESCRIPTION DETAIL
BlueData EPIC 2.0 Software Infrastructure software that supports virtual ‘compute’ clusters of CDH, HDP and Spark
stand-alone.
CDH clusters can be configured with Cloudera Manager (and Cloudera Navigator) while HDP
clusters can be configured with Ambari (and other services enabled by Ambari)
Another key feature is DataTap that provides a HDFS protocol URI dtap:// that allows virtual
‘compute’ clusters running on BlueData to leverage external storage systems such as Isilon
via HDFS or NFS.
Note: CDH and HDP clusters are **NOT** configured with Isilon directly. These clusters
connect to Isilon via BlueData DataTap functionality
RELEVANT EMC FEATURES
Table 2 List of EMC technology features relevant to the validation consideration
DESCRIPTION DETAIL
EMC Isilon HDFS Module, NFS
VALIDATION SCOPE
PARTNER TECHNOLOGIES
Table 3 List of Partner technologies used in the validation process
TECHNOLOGY - MODEL TECHNOLOGY VERSION TECHNOLOGY INSTANCE (PHYSICAL/VIRTUAL)
BlueData EPIC EPIC 2.0 Controller Node: Physical
Worker Node: Physical
Cloudera virtual Hadoop cluster CDH 5.4.3 with Cloudera
Manager Enterprise
1 Master Node: Virtual (Docker Container)
3 Worker Node: Virtual (Docker Container
Hortonworks virtual Hadoop cluster HDP 2.3 with Ambari 2.1 1 Master Node: Virtual (Docker Container)
3 Worker Node: Virtual (Docker Container)
EMC TECHNOLOGIES
Table 4 List of EMC technologies used in the validation process
TECHNOLOGY - MODEL TECHNOLOGY VERSION TECHNOLOGY INSTANCE (PHYSICAL/VIRTUAL)
Isilon X210 Isilon OneFS v7.2.1.0
B_7_2_1_014(RELEASE)
Physical (8 node system)
Data Access Methods
Table 5 List of data access methods and technologies used in the validation process
DATA ACCESS METHOD DATA ACCESS STRUCTURE DETAILS
Isilon HDFS API Over the wire HDFS protocol implemented in Isilon
TESTING ENVIRONMENT
ARCHITECTURE ILLUSTRATION

EMC Corporation 8
Figure 1. Illustration of logical layout of Partner and EMC technologies used for validation
RESOURCES
Table 6 List of resources and technologies used to support the validation process
DESCRIPTION VERSION DETAILS & SPECIFICATIONS
HARDWARE
(6) BlueData Workers SuperMicro model: X9DRFR
6 workers: SuperMicro, 10G NIC, 64GB RAM, 4 cores, 8 cpu, 5 1TB drives
1 controller: Same as the worker, but with 128GB ram
CPU: Intel(R) Xeon(R) CPU E5-2609 v2 @ 2.50GHz
All boxes have the same 5 drives:
1TB Seagate SSDH hybrid drives (ssd + disk) sn: Z1DG4XE
(1) BlueData Controller SuperMicro model: X9DRFR
(8) Isilon X210 8 node system
CDH Virtual Cluster Functional testing and validations
(1) Master Node Master: XXL - 8 VCPU, 24576 MB RAM
(6) Slave Nodes Worker: Extra Large - 4 VCPU, 12288 MB RAM

EMC Corporation 9
BLUEDATA EPIC DOCUMENTATION
1. About BlueData EPIC
2. Installation Guide
3. User-Administrator Guide
BLUEDATA INSTALLATION AND CONFIGURATION
Note: There are various pre-requisites for BlueData EPIC Software. These are related to Hosts, Operating System, Networking and
High Availability. Details can be found here
Phase One: Installing the BlueData EPIC Controller
To install EPIC from the command line:  
1. Install Red Hat Enterprise Linux 6.5 or CentOS 6.5 on the hosts that you will use for the Controller and Worker nodes.  
2. Log into the host that you will be using as the Controller node using the root account and password.  
3. Download the EPIC Enterprise binary (.bin) from BlueData Software, Inc. to the host that you will use as the Controller
node. The size of the download will depend on the distribution(s) included and the flavor of the .bin file.  
4. Make the .bin file executable by executing the command
5. chmod a+x bluedata-epic-<type>-<os>-<package>-release-<version>-<build>.bin
Where:
- <type> can be either entdoc (for EPIC) or onedoc (for EPIC Lite)  
- <os> is the operating system supported by this .bin file. This can be either centos (for CentOS) or rhel
(for Red Hat Enterprise Linux).  
- <package> is the EPIC package type. This will be either eval (EPIC Lite), full (CentOS with all App Store
images and OS packages included), or minimalplus (RHEL with all App Store images included).  
- <version> is the EPIC version being downloaded.  
- <build> is the specific EPIC build number being   downloaded.   5. Run the executable binary from the
Linux console as the root user by typing ./<epic>.bin --floating-ip <address/range>, where:  
- <epic> is the full name of the .bin file (see Step 4, above).  
- <address/range> is the CIDR to use, such as 192.168.25.10/24. This range of addresses allows
network access from outside the EPIC platform to the virtual nodes that EPIC will create as part of future clusters.
The installer checks the integrity of the EPIC bundle and then extracts the bundle contents and performs some initial checks.
The installer checks to see which network interface card(s) (NIC) have Internet connectivity.
- If multiple NICs can connect to the Internet, then a prompt appears asking you to select the NIC to use.  
- If only one NIC can connect to the Internet, then the installer bypasses this step.  
- If no NICs can connect to the Internet, the installer asks for the IP address of the proxy server being used for your network.   If
prompted, enter either the number corresponding to the NIC to use to connect to the Internet or the IP address of the proxy server,
and then press [ENTER].  
The End User License Agreement (EULA) appears. Read through the EULA, pressing [SPACE] to page through the content. Once you
have viewed the entire EULA, press y to accept it and continue installing EPIC.

EMC Corporation 10
EPIC installs on the Controller node. A series of messages appear during the installation. The following message appears once the
installation is complete:
Figure 2.1: Installation complete
This concludes the first phase of the EPIC installation. Note the URL provided, as you will use this to continue configuring EPIC.
Phase Two: Completing the Installation
The next step of the EPIC installation process is to configure EPIC using a Web browser to access the application. To do this:
1. Open a Web browser and navigate to the URL provided at the end of the command line installation process.
The EPIC Enterprise - Setup screen appears.
3. The read-only Public network interface field displays the NIC that you selected for Internet access during the command line
installation. Each host in the EPIC platform must use the same NIC to access the Internet. For example, if you selected the eth0 NIC
on the Controller node, then the eth0 NIC on each Worker node must also be able to reach the Internet.
4. Use the System Storage pull-down menu to select the type of system storage to use for the EPIC platform. The available options
are:
If the nodes each have a second or third hard drive as described in the About EPIC Guide and you want to create local
system storage using HDFS with Kerberos protections, then select Create HDFS from local disks for system storage.  
. To use an existing external HDFS file system as system storage, select Use existing HDFS for System Storage and then
enter the parameters for “HDFS”
. To use an existing external NFS file system as system storage, select Use existing NFS for System Storage and then enter
the parameters described for “NFS”  
Note: Configuring system storage does not preclude the ability for clusters running on BlueData to reach out to other storage
systems using DataTap. The system storage is just a default scratch space for each tenant and is referred to as Tenant Storage.
If you are creating local HDFS system storage, then select one or more hard drive(s) to use for this storage in the Select one or
more available disk(s) for HDFS field.  
Click Submit to finish installing EPIC on the Controller node. EPIC displays a popup indicating that the installation process has

EMC Corporation 11
  started successfully.   Figure 2.3: Installation Started popup  
This popup is soon replaced by a status summary as the installation completes. If you like, you may click the green Details button to
open a popup that displays additional information about the installation. Please allow about 20 minutes for this process to complete
(actual time will vary depending on various factors).
Figure 2.4: Completing the EPIC installation
The Bluedata software setup completed successfully popup appears when the installation process is completed. Click the Close
button to exit to the EPIC Login screen,
Figure 2.5: Installation completed
Proceed to the next section to begin adding Worker nodes.
Phase Three: Adding Worker Nodes
Once you have finished installing EPIC on the Controller node, the final step is to add the Worker nodes to the EPIC platform. As
such, workers provide additional CPU and memory so that you can create more virtual nodes. Each Worker node or host must
conform to the system requirements listed in the About EPIC Guide and any applicable Deployment Guide.
Note: This section provides a high-level overview intended to help you get up and running with EPIC as quickly as pos- sible. Please
see the User/Admin Guide for information about this screen and any applicable Deployment Guide for further directions.

EMC Corporation 12
To add Worker nodes:
1. Access the EPIC Login screen by opening a Web browser and navigating to the Controller IP address.
2. Enter your username and password in the appropriate fields and then click the red Login button.
3. In the main menu, select Installation. The Cluster Installation screen appears.
4. Enter the IP address(es) of the Worker node(s) that you wish to add to the EPIC platform in the Worker IP field. You may enter IP
addresses as follows, being sure not to add any spaces:
- Single IP address: Enter a properly formatted IP address, such as 10.10.1.1.  
- Multiple IP addresses: Enter the first three octets of the IP addresses, and then separate each digit of the fourth octet
with a commas, such as 10.10.1.1,2,5,8. In this example, EPIC will install four Worker nodes with IP addresses of 10.10.1.1,
10.10.1.2, 10.10.1.5, and 10.10.1.8.
- Multiple IP addresses: Enter multiple IP addresses separated by commas, such as

EMC Corporation 13
10.10.1.1,10.10.1.2,10.10.1.5,10.10.1.8. In this example, EPIC  will install four Worker nodes with the same IP addresses as the
previous example.  
- IP address range: Enter an IP address range, such as 10.10.1.1-8. In this example, EPIC will install eight Worker nodes
with IP addresses from 10.10.1.1 to 10.10.1.8.  
- Combination: Use a combination of the above methods, such as 10.10.1.1,10.10.1.2,5,8,10.10.1.9-12.  
5. Select how to access the Worker node(s). Your available options are:
cCicking the Private Key field to open a standard File Upload dialog that allows you to browse for and select the key file. If the key
requires a pass phrase, enter that phrase in the Passphrase field. The uploaded private key will only be used for initial access to the
worker nodes and will not be permanently stored.
Root access: Check the Root Access radio button and then enter the root password for the Worker node(s) you are installing in the
Root Password field..
Figure 2.8: Root Access information
SSH Key: If the Worker node(s) have a public key installed to allow password-free access, then you may check the SSH Key based
Access radio button. Upload the private key by
Figure 2.9: SSH Key information
6. Click the blue Add Workers button to install the selected Worker node(s).
EPIC will prepare the selected Worker node(s). A series of colored indicator bars in the Worker(s) Status table will display the
installation progress for each Worker node, as follows:
Figure 2.10: Worker node installation progress

EMC Corporation 14
-
Once you have finished adding Worker node(s), you may view the status of each node by clicking Dashboard in the main menu to
open the Dashboard screen, and then selecting the Services tab, which presents detailed status information for each Worker node.
Note: You may only perform one set of Worker node instal- lations to one or more node(s) at once. To save time, con- sider adding
all of the Worker nodes at once by entering multiple IP addresses as described above.
Note: Verify that the Worker node(s) have finished rebooting before attempting to create a virtual cluster. Check the Ser- vice Status
tab of the Dashboard screen to ensure that all services are green before proceeding.
Figure 2.15: Site Admin Dashboard screen - Service Status tab
Once the installation of the workers is complete, you can login to BlueData UI, change to Demo Tenant and create Hadoop or Spark
clusters using this simple one page form. BlueData automatically provisions the virtual nodes (Docker containers); does the
networking and isolation; deploys the software for the relevant distribution and configures the applications (YARN, M/R, Hive etc).

EMC Corporation 15
EMC ISILON INSTALLATION AND CONFIGURATION
Install
The physical assembly and configuration of the Isilon cluster was performed without incident using some basic information provided
by email from EMC.
Screenshots of EMC Isilon Set Up

EMC Corporation 16
HDFS with Kerberos Authorization
The creation of the necessary Kerberos Authentication Provider, SPNs, and proxyusers on the Isilon was performed without incident
after following the instructions in the OneFS-7.2.0-CLI-Administration-Guide and the Configuring Kerberos with HWX and Isilon white
paper.

EMC Corporation 17
Screenshot of EPIC DataTap Configuration Screen (to Kerberized Isilon HDFS)

EMC Corporation 19
Isilon HDFS Access Zone configuration
Isilon HDFS Proxyusers Configuration
NFS with Simple Authorization

EMC Corporation 20
The creation of an NFS export and associating it with the System Access Zone was performed without incident. The only issue
encountered during configuration was that the default setting of the NFS export in Isilon which mapped the “root” user to “nobody”
had to be changed in order to permit the “root” user specified by the DataTap to write to the NFS share.
Screenshot of EPIC DataTap Screen (NFS DataTap)
HDFS with Simple Authorization
The creation of the Local Authentication Provider with “bluedata”, “hdfs”, and “root” members and configuring it with the System
Access Zone was performed without incident.
During basic connectivity testing between DataTap and Isilon using the HDFS protocol, an issue with the DataTap issuing of
commands to the DataNode was uncovered. Commands issued by DataTap to the DataNode were returned with a “Lease expired”
error. The root issue was that for the Isilon HDFS implementation the contents of the “clientname” fields had to be kept consistent
across NameNode and DataNode request packets. This restriction was not observed when running DataTap against a native Apache
HDFS instance. The EPIC DataTap code was changed to maintain this clientname consistency.
Screenshot of EPIC DataTap Screen (HDFS Non-Kerberized DataTap)

EMC Corporation 21
Installation of BlueData EPIC was performed using the standard installation procedure. There were no Isilon specific dependencies.
SUPPORTING INFORMATION AND CONSIDERATIONS
The Table below contains pertinent information relating to supporting the testing environment.
Table 7 List of supporting information and considerations relevant to the validation process
DESCRIPTION DETAIL
Isilon Kerberos Setup
BlueData Installation Guide This is an EMC-internal application for automating the execution of benchmarks and
collecting performance metrics.
VALIDATION TESTS
COMPATIBILITY TESTING
TEST # 1: Connect to Kerberized Isilon HDFS via BlueData DataTap object.
a. Test Goal: Ability to connect to remote Kerberized Isilon HDFS
b. Pass Criteria:
i. Browse the directories and files in the Kerberized HDFS Access Zone with trusted user
ii. Permissions denied message if un-authorized users attempt access to Kerberized Isilon HDFS

EMC Corporation 22
c. Test Preparation:
1. Set up standalone KDC
2. Configure HDFS Access Zone with KDC
3. Generate keytab file for a specific principal (hdfs)
d. Test Plan:
1. Login to BlueData tenant
2. Create new DataTap of type = HDFS
3. Specify Kerberos credentials including keytab file for the relevant user (see screenshot above in
installation and configuration section)
4. Click submit
5. Click on the DataTap name to browse the data in Isilon – should match all the directories/folders
as seen in Isilon file system browser
e. Result: PASS (see screenshots below)

EMC Corporation 23
Isilon log showing non-proxyuser being denied DataTap Access
EPIC Running a CDH 5.4.3 virtual cluster returns an error when un-authorized user “llma” attempts to access an Isilon HDFS file
system protected by Kerberos authorization via a DataTap. Access to the same HDFS file system via DataTap succeeds when
performed by authorized user “bluedata”.

EMC Corporation 24
TEST # 2: Connect to Isilon via BlueData NFS DataTap
a. Test Goal: Ability to connect to remote Isilon via NFS protocol
b. Pass Criteria: Browse the directories and files in the NFS Access Zone
1. Configure NFS access zone in Isilon
d. Test Plan:
2. Create new DataTap of type = NFS
3. Click submit
4. Click on the DataTap name to browse the data in Isilon – should match all the directories/folders
as seen in Isilon file system browser
e. Result: PASS (see screenshots below);

EMC Corporation 25
TEST # 3: Browse Kerberized Isilon HDFS DataTap from CDH (Cloudera) Compute
Cluster
a. Test Goal: Ability to connect to Kerberized Isilon HDFS DataTap from any Hadoop compute cluster without any
specialized configuration (i.e. without any distribution specific integration with Isilon HDFS)
b. Pass Criteria: Login to master node of Hadoop cluster and issue hadoop fs commands against Isilon HDFS DataTap
1. Create a Cloudera (CDH) virtual cluster using BlueData cluster
CDH virtual
Utilize the Isilon DataTap (isi-hdfs-kerb)

EMC Corporation 26
d. Test Plan:
1. ssh into master node (10.39.250.2) of virtual Hadoop cluster (see above)
2. Execute command: hadoop fs – ls dtap://isi-hdfs-kerb
3. The directories/folders listed must match Isilon file system browser

EMC Corporation 27
TEST # 4: Browse Kerberized Isilon HDFS DataTap from HDP (Hortonworks)
Compute Cluster
f. Test Goal: Ability to connect to Kerberized Isilon HDFS DataTap from any Hadoop compute cluster without any
specialized configuration (i.e. without any distribution specific integration with Isilon HDFS such as using Cloudera
Manager or Ambari)
g. Pass Criteria: Issue hadoop fs commands against Isilon HDFS via BlueData DataTap URI (dtap://)
h. Test Preparation:
1. Create a Hortonworks (HDP) virtual cluster using BlueData

EMC Corporation 28
Note: The IP address of the virtual master node of the Hadoop cluster is 10.39.250.6
i. Test Plan:
1. ssh into master node of virtual cluster
2. Execute command: hadoop fs – ls dtap://isi-hdfs-kerb
3. The directories/folders listed must match Isilon file system browser
j. Result: PASS (see screenshots below);
TEST # 5: Run a Hive query against data stored in Isilon Kerberized HDFS via Hue
console
k. Test Goal: Run Hadoop jobs directly against data in-place on Isilon HDFS
l. Pass Criteria: Create external table definitions against data stored in HDFS and obtain results from query
m. Test Preparation:

EMC Corporation 29
1. Create a CDH (Cloudera) virtual cluster using BlueData
2. Login to Hue
n. Test Plan:
1. Create external table ‘ratings’ against ratings.csv stored on Isilon HDFS under folder = ratings
2. Create external table ‘products’ against geoplaces2.csv stored on Isilon HDFS under folder =
products

EMC Corporation 30
3. Create a join query between ratings and products tables
4. Execute the query
o. Result: PASS

EMC Corporation 31
TEST # 5: Integrate with Cloudera Navigator for metadata search and auditing
a. Test Goal: Confirm that Cloudera Navigator can audit and track the metadata (incl. lineage) of compute services.
b. Pass Criteria: Cloudera Navigator search shows all the jobs from the compute cluster against Isilon
1. Create CDH 5.4.3 cluster
2. Use the Cloudera Manager (that is instantiated by BlueData) to configure Cloudera Navigator for
that above cluster (note that Cloudera manager and hue console are running on 10.35.231.2).
Cloudera Navigator services are configured to run on 10.35.231.3

EMC Corporation 32
3. Use the IsilonHDFS DataTap - note the configuration below:
d. Test Plan:
2. Use the BlueData console to a Hadoop M/R job utilizing the aforementioned cluster against data
stored in Isilon (i.e. via dtap://)

EMC Corporation 34
3. Use the Hue Console to create external tables & execute queries against data stored in Isilon and
run a join query. Note the execution time stamps
4. Login to Cloudera Navigator – confirm that the metadata search and audit logs show each of the
actions as well as sub operations with the appropriate time stamps. Testing was specifically done
with HIVE, IMPALA & M/R

EMC Corporation 35
M/R Job
The time stamp of the job from the test preparation screen is: Oct 31, 2015 at 8:15:43
Note that Navigator has the same time stamp and the job input and job output to IsilonHDFS datatap is also shown here.
HIVE QUERY: Note the time stamp from the previous Hue screens of 10/30/2015 from 9.51am
Note that the external table is using the ratings folder in Isilon

EMC Corporation 36
Note the corresponding audit events in Cloudera Navigator for HIVE
IMPALA: Note the time stamp in Hue of 10/31/2015, 8.04am

EMC Corporation 37
TEST # 6: Preliminary performance benchmarks using standard development/test
hardware in BlueData Labs.
a. Test Goal: Run the Intel HiBench-Terasort test suite using BlueData compute clusters against various data sizes in
Isilon
b. Pass Criteria: Successful completion of all the terasort for 100GB and 1TB and measurement of execution times in
seconds.
i. Note: Tuning performance was a non-goal since this environment used 1 Gig network vs. the
recommended 10 gig network
ii. This size of VMs used were limited to 51GB RAM due to the specs of the underlying hardware (dev/test
machines)
1. Specs of Hardware for running BlueData compute cluster:
Hosts:
SuperMicro model: X9DRFR
Switch:
Arista 10G switch, model:
DCS-7050T-36 32x 10GBASE-T + 4x SFP+
2. Create special VM flavors for master/worker nodes
Master: benchmark - 6 VCPU, 51200 MB RAM, 350 GB root disk
Worker: benchmark - 6 VCPU, 51200 MB RAM, 350 GB root disk
3. Create CDH 5.4.3 cluster

EMC Corporation 38
4. Create Isilon DataTap (isi-hdfs)

EMC Corporation 39
d. Test Plan:
2. Use the BlueData console to a Hadoop M/R job utilizing the aforementioned cluster against data
stored in Isilon (i.e. via dtap://)
3. Use the Hue Console to create external tables against data stored in Isilon and run a join query
4. Login to Cloudera Navigator – confirm that the metadata search and audit logs show each of the
actions as well as sub operations.
5. SEE APPENDIX SECTION FOR THE ACTUAL COMMANDS EXECUTED
e. Result: PASS
TERASORT 100GB 2909.479 seconds
TERASORT 1TB 33887.223 seconds

EMC Corporation 41
TESTING ENGINEERS
NAME TITLE COMPANY PHONE EMAIL
Boni Bruno Principal Solutions
Architect
EMC 818-297-4571 Boni.Bruno@emc.com
Anant Chintamaneni VP, Products BlueData 650-906-7312 anant@bluedata.com
Tom Phelan Co-Founder BlueData tap@bluedata.com
TESTING ACCEPTANCE
NAME TITLE COMPANY PHONE EMAIL
Boni Bruno Principal Solutions
Architect
EMC 818-297-4571 Boni.Bruno@emc.com
Anant Chintamaneni VP, Products BlueData 650-906-7312 anant@bluedata.com
Tom Phelan Co-Founder BlueData tap@bluedata.com

EMC Corporation 42
APPENDIX
Teragen command

EMC Corporation 43
Terasort command

EMC Corporatoin 45
Teravalidate command

EMC Corporation 46
TEST # 7: Hadoop Cluster (CDH) with Isilon as HDFS replacement
Cloudera, on standalone infrastructure was configured directly with Isilon to understand the value proposition of BlueData as the
infrastructure layer with DataTap.
Key Observations:
1. Manual deployment of Cloudera using Cloudera Manager and Isilon as services option required a custom workflow.
Assumption is that each subsequent CDH cluster will require the same workflow  BlueData offers a streamlined
workflow. Cloudera compute clusters (e.g. different versions or multiple) can be created unmodified without
any dependency on Isilon.
2. Cloudera Navigator Lineage (metadata search) for compute services do not function in the Cloudera+Isilon deployment 
BlueData supports lineage for all compute services. BlueData also does not support HDFS level auditing (that
Cloudera Navigator supports with DAS clusters)
Cloudera Standalone (without BlueData) + Isilon Screenshot
Cloudera Manager running on 10.35.248.5

EMC Corporation 47
Isilon configuration in Cloudera Manager
Default settings were used. Specific access zone was created in Isilon (see next screenshot)
Isilon Networking and Access Zone
Cloudera Standalone (without BlueData) Cluster Details

EMC Corporation 48
Comparison of Cloudera on BlueData + Isilon vs. Cloudera Standalone + Isilon (Navigator)
Since BlueData abstracts the HDFS storage, it does not currently support HDFS auditing. However, BlueData supports auditing and
metadata lineage of all compute services. Cloudera Standalone deployed with Isilon does not support metadata lineage in Navigator.
The table below summarizes these key observations for key use cases
No Use case Cloudera on
Bluedata + Isilon
Cloudera Standalone
+ Isilon
1 Metadata – Hive Query YES NO
2 Metadata – Impala Query YES NO
3 Metadata - Pig YES NO
4 Metadata - M/R YES NO
5 Metadata - Oozie YES NO
Based on the testing pattern, the key compute lineage/metadata operations are NOT
captured by Navigator in the case of Cloudera standalone + Isilon.
1. Metadata – Hive Query
Test Preparation:
On both environments, create a simple table ‘ratings’ and run a group by query as shown below

EMC Corporation 49
Cloudera Standalone + Isilon (10.35.248.8); Note the time stamps of 12/07/2015
11:17pm
Cloudera Standalone + Isilon
(note the time stamp on Mac); No results are logged by Navigator

EMC Corporation 50
Cloudera on BlueData + Isilon (10.35.231.2);
Note the time stamp of 12/07/2015 11:21pm
Cloudera on BlueData + Isilon
(note the time stamp on Mac ); Both Hive and YARN operation executions
are logged by Navigator. Note that the time stamp matches the time stamps
in the Hue console for the corresponding query.

EMC Corporation 51
2. Metadata – Impala Query
(note the time stamp on Mac); No results are logged by Navigator
(note the time stamp on Mac); Impala operation execution is logged with the same time
stamp as in Hue

EMC Corporation 52
3. Metadata – Pig Query
(Note the time stamp for the Pig Job. No results captured in Navigator)

EMC Corporation 53
(Note the time stamp for the Pig Job. Results are captured in Navigator)

EMC Corporation 54
4. Metadata – Oozie
(Note the time stamp for the Oozie Job. No results captured in Navigator)

EMC Corporation 55
(Note the time stamp and ID for the Oozie Job. Results are captured in Navigator)

BlueData Isilon Validation Brief

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to BlueData Isilon Validation Brief (20)

BlueData Isilon Validation Brief