HPE ConvergedSystem 300 for Microsoft Analytics Platform: Data

HPE ConvergedSystem 300 for
Microsoft Analytics Platform
Data Integration Platform – Gen9 reference guide for
the AU5 release
Technical white paper
Technical white paper
Contents
Executive summary .................................................................................................................................................................................................................................................................................................................................3
Overview of services for the Data Integration Platform.........................................................................................................................................................................................................................................3
Updates ..............................................................................................................................................................................................................................................................................................................................................................3
Solution overview ..................................................................................................................................................................................................................................................................................................................................... 4
Backup and restore platform .................................................................................................................................................................................................................................................................................................. 4
Data Loading Staging Platform ............................................................................................................................................................................................................................................................................................ 5
Measured performance ...................................................................................................................................................................................................................................................................................................................... 6
Platform configuration .................................................................................................................................................................................................................................................................................................................. 6
Database configuration ................................................................................................................................................................................................................................................................................................................ 6
Test details .............................................................................................................................................................................................................................................................................................................................................. 7
Performance results ........................................................................................................................................................................................................................................................................................................................ 7
Licensing and support considerations ................................................................................................................................................................................................................................................................................. 8
Licensing considerations ............................................................................................................................................................................................................................................................................................................8
Support considerations ................................................................................................................................................................................................................................................................................................................ 8
General considerations ....................................................................................................................................................................................................................................................................................................................... 8
Recommended hardware for the Data Integration Platform ..................................................................................................................................................................................................................... 8
Recommended networking configurations for the Data Integration Platform.......................................................................................................................................................................... 9
Recommended software for the Data Integration Platform ..................................................................................................................................................................................................................... 12
Caveats ............................................................................................................................................................................................................................................................................................................................................................ 12
Appendix A – Recommended configuration for a Data Integration Platform (with part numbers) .......................................................................................................................... 13
Resources and additional links .................................................................................................................................................................................................................................................................................................. 18
Technical white paper
Page 3
Executive summary
HPE ConvergedSystem 300 for Microsoft Analytics Platform addresses the key needs of business insights through greater understanding of
business data. It is pre-configured and the only purpose-built solution customized for Data Warehouse (DW) with the HPE Support Pack to meet
a range of business needs. The HPE ConvergedSystem 300 for Microsoft Analytics Platform can be easily and rapidly deployed to support both
large relational database solutions with Microsoft® SQL Server Parallel Data Warehouse. All required components, including servers, storage, and
networking, are integrated into a single ConvergedSystem that is expandable to meet virtually any big data challenge.
The HPE ConvergedSystem 300 for Microsoft Analytics Platform comes as a self-contained solution. However, there are recommended functions
that can be supported with a separate server (or set of servers) to provide optimal backup, restore, and data loading (Extract, Transform, and
Load). This paper will review these external capabilities and make configuration recommendations.
It is important that the reader have a good understanding of the HPE ConvergedSystem 300 for Microsoft Analytics Platform solution prior to
reading this paper. It is strongly recommended that the reader review the HPE ConvergedSystem 300 for Microsoft Analytics Platform solution
architecture white paper or similar documentation before proceeding.
Target audience: The target audience for this document is IT directors, IT architects, database administrators, business intelligence (BI)
administrators, and others wishing to learn more about extending the functionality of the HPE ConvergedSystem 300 for Microsoft Analytics
Platform solution. This document assumes the reader has a basic understanding of several key data center technologies including, but not
limited to: servers, storage, networking, power, solution management, virtualization, and hypervisors.
Overview of services for the Data Integration Platform
The Data Integration Platform (DIP) refers to one or more servers you may elect to accompany your purchase of an HPE ConvergedSystem 300
for Microsoft Analytics Platform solution. You may also choose to re-purpose an existing server or set of servers in your environment to provide
this functionality. The Data Integration Platform server(s) will provide auxiliary support for your system, including support for the following
capabilities:
• Intermediate location for backups (and therefore a staging area for restore operations)
• A staging location for Extract, Transform, and Load (ETL) or ELT operations
The Data Integration Platform must be ordered separately from your HPE ConvergedSystem 300 for Microsoft Analytics Platform solution.
Updates
This Data Integration Platform continues to be based on the HPE ProLiant DL380 Gen9 platform, and now features the Intel® Xeon® E5-2600
Version 4 processors.
The HPE ProLiant Gen9 portfolio is a major delivery milestone in HPE’s compute strategy, which addresses IT demands with a vast pool of
processing resources that can be located anywhere, scaled to any workload and available at all times. The servers are optimized for convergence,
cloud and software-defined environments, and feature new technology innovations such as:
• HPE-unique PCIe accelerators and 2400MHz DDR4 SmartMemory that increase compute capacity
• Converged management across servers, storage and networking to enable a software defined enterprise
• Faster setup, monitoring and firmware maintenance with reliable, secure and innovative embedded management, including UEFI and RESTful
APIs for hybrid cloud environments
• PCIe Accelerators, HPE SmartCache and HPE FlexFabric adapters that deliver improved performance
Technical white paper
Page 4
Solution overview
The HPE DIP server is a companion product for the HPE ConvergedSystem 300 for Microsoft Analytics Platform that is purchased separately.
The DIP server can be customized to fit each customer environment and server configuration. This document provides a tested reference guide
for CS300 appliance Backup/Restore/ and Load. The DIP server, while not part of the appliance, can be racked into the appliance rack (if space
allows) and is designed to connect to the appliance network switches. This section details the tested configuration for this reference guide.
HPE ConvergedSystem 300 for
Microsoft Analytics Platform
Data Integration Platform
HPE ProLiant DL380 Gen9
Figure 1. Overview
For additional product information you can read the QuickSpecs available at:
http://h20195.www2.hpe.com/V2/GetDocument.aspx?docname=c04126041
Backup and restore platform
A key requirement of most database and analytics data storage systems is to have a backup of the data stored on the server. This backup
location is typically going to be migrated off-site for extended storage and disaster recovery purposes. However, in the interim, a local copy of the
backup is taken for efficiency and performance of the backup itself. The Data Integration Platform is the ideal location as the first storage location
for Microsoft Analytics Platform System (APS) backups.
Technical white paper
Page 5
SQL Server Parallel Data Warehouse backup
SQL Server Parallel Data Warehouse (PDW) offers two backup types for your PDW databases. You can either take a full database backup (as it
sounds, a copy of everything), or a differential backup, which will contain everything in your database that has changed since the last time a full
database backup was taken. A differential backup will depend upon the last full backup for its baseline. For further details, review the APS.chm
documentation file.
For each of these backup types, SQL Server Parallel Data Warehouse produces a set of files. These files together, rather than a single file, make
up the database backup or differential backup. These backups will always use compression. The BACKUP DATABASE syntax assumes that the
backup location will be on disk, via a UNC path to a specific directory to retain the backup (and must NOT be on the system itself). Therefore, a
separate backup server is needed, with enough disk storage space to contain the full database backups and differential backups for each user
database. Additionally, as with any local backup strategy an off-site or enterprise backup solution should be employed to make secondary
(offsite) backups.
Note
Backup sets cannot be restored to a smaller system, so if a disaster recovery strategy involves using backups, the other system being restored to,
must have at least the same number of compute nodes.
You should also back up the master database of the control node. Note, however, that a backup of the master database is a regular SQL Server
backup of the master database, and must always be a full database backup. The master database will likely not be significant in size, but is an
integral part of any recovery operation. It is strongly recommended that the master database backups use a parallel strategy to user database
backups in terms of off-site storage, DR recovery planning, etc.
SQL Server Parallel Data Warehouse restore
To restore any individual database, a copy of the last full database backup must be available at an accessible UNC location. Once the full
database backup is applied, the latest differential backup can then be applied to bring the database up to the latest point of recoverability. The
last differential backup must also be accessible via a UNC share for the restore process.
If the entire system must be restored, the restore of the master database on the control node should be performed, and then the recovery as
noted above for each individual user database (via the full and latest differential backup). Each of these backup file sets should be present on the
Data Integration Platform for optimal recovery performance.
Using the Data Integration Platform
The recommendation for the backup and restore platform is that the server should be connected to the solution on the InfiniBand network if
possible, to maximize the performance of the data movement, and use 10Gb Ethernet to connect to the customer corporate network. The Data
Integration Platform as documented in this paper meets these requirements and will provide optimal networking and data transfer capabilities for
the solution.
Data archival
It is strongly recommended that the database backups will be copied to a remote storage location (for example, a Disaster Recovery data center).
The Data Integration Platform is an ideal location to have the “latest” backups (for faster recovery purposes), but typically does not have enough
storage to have multiple copies of database backups. Additionally, for disaster recovery planning, a remote copy of the data is critical in the event
of a severe event affecting the primary data center.
Data Loading Staging Platform
SQL Server Parallel Database Warehouse data will typically be loaded into the appropriate database using one of the following mechanisms:
• ETL Software solution, such as SQL Server Integration Services (SSIS)
• DWLoader utility (provided with the HPE ConvergedSystem 300 for Microsoft Analytics Platform solution)
• A custom data loading process incorporating one or more of the above tools
• A third party ETL solution
Technical white paper
Page 6
For most usages of data loading, an interim platform must be used to stage the data before it can be loaded into the SQL Server PDW database.
Assuming, as above, that the Data Integration Platform is connected to the solution via the InfiniBand network, data can be transferred and
loaded using the high bandwidth InfiniBand interconnects.
The Data Integration Platform is also an ideal location for any workflow or transformation engine(s) needed to support the data loading activities.
Note that most data loading activities will not necessarily require significant CPU scalability, but the ETL software may be more CPU intensive.
If being used as a data staging location, focus on the storage capabilities and capacity of the Data Integration Platform storage system. If you are
also planning on hosting your ETL engine (such as SSIS) on the server, consider adding additional CPU and memory capacity to the server.
Measured performance
Starting with the AU3 release, performance baselines have been created for the DIP server platform. These baselines will be used to compare
performance on future releases of the appliance and the DIP platform. This section details the performance tests that were run as part of the
baseline creation.
Platform configuration
Several different platform configurations were used for the performance testing, from 2, 4, 8 and 16 node systems running on HPE Generation 9
(Gen9) hardware with the AU4 software to an 8 node system running with the recently released AU5 software.
16 Node system – The 16 node system used 2 full racks with 16 compute nodes to test a multi-rack configuration.
8 Node system – The 8 node system was comprised of 8 compute nodes, and was used to establish a full rack performance baseline.
4 Node system – The 4 node system tested a ½ rack configuration.
2 Node system – The 2 node system used 2 compute nodes to create a quarter-rack baseline for Gen9/AU4.
Note
8 Node configurations and larger, now feature a “High Performance” drive option that substitutes 15K SAS drives for the standard SAS MDL
drives. These drives support faster transfer rates at the expense of lower storage density.
Database configuration
For the SQL server performance tests, a 1TB TPCh database was created on the CS300 APS appliance. This database was set up to use both
distributed and replicated tables. For tests that use indexes, indexes were created on the tables using clustered columnstore. For the SQL table
scan throughput test, the indexes were deleted. Table 1 below details the settings for the various TPCh database tables.
Table 1. Database Setup
TABLE
DISTRIBUTED
OR REPLICATED
INDEX
Lineitem
Distributed
Clustered/Columnstore
Orders
Distributed
Clustered/Columnstore
Partsupp
Distributed
Part
Replicated
Supplier
Replicated
Customer
Replicated
Region
Replicated
Nation
Replicated
Technical white paper
Page 7
Note
The test performed for this performance baseline DO NOT constitute a formal TPCh benchmark. The TPCh database was used to provide test
data only.
Test details
Five different tests were performed for the performance baseline. These tests include SQL Server DB administration tasks as well as tests to
measure throughout both in RAW form and through SQL table scans.
• Database load rate: This test measures the aggregate load performance of the DIP server and the appliance. For this test, unindexed heap
tables were loaded simultaneously through both InfiniBand connections. The flat data files are created and stored on the DIP server.
• Database backup rate: Overall Backup rate is measured in this test. The 1TB database is backed up through a single InfiniBand connection to
a file share on the DIP server. The Backup set size is divided by the time taken to back up the Database to determine the backup rate.
• Database restore rate: The Database restore rate is measured by taking the previous backup and restoring it through a single InfiniBand
connection. The Backup set size is then divided by the time taken for the restore to complete.
• SQL Server scan rate: SQL Scan rate is measured by performing a table scan on a large table. In this case, the index on the lineitem table is
deleted, and a “Select count_big (*) from lineitem” is performed to scan the table and count the records. Performance monitor is used to
aggregate the I/O rates on each of the 16 LUNs on each HSA storage node.
• RAW I/O throughput: This is a non-SQL server test using the Microsoft SQLIO utility to measure raw I/O performance on all LUNs within a
scale unit. A data file is created on each LUN, and this file is then read using connection parameters to determine the maximum I/O rate on
each scale unit.
Performance results
The performance results are detailed in Table 2.
Table 2. Performance Metrics
PERFORMANCE METRIC
HPE 16
NODE
HPE 8 NODE
(AU5)
HPE 4
NODE
HPE 2
NODE
COMMENTS
Database Load Rate
(High Performance)
5.2TB/Hr
7.1TB/Hr
5.1TB/Hr
7.0TB/Hr
4TB/Hr
N/A
2.1TB/Hr
N/A
Loading TPCh data into 8 heap tables concurrently over 2 IB
connections
Database Backup Rate
5.1TB/Hr
8.4TB/Hr
5.0TB/Hr
6.6TB/Hr
4.4TB/Hr
N/A
2TB/Hr
(High Performance)
4.5TB/Hr
6.6TB/Hr
4.0TB/Hr
5.0TB/Hr
3.7TB/Hr
N/A
366GB/Hr
N/A
SQL Server Scan Rate
63GB/Sec
36.8GB/Sec
18.4GB/Sec
9.2GB/Sec
RAW I/O Throughput
67.6GB/Sec
38.8GB/Sec
19.4GB/Sec
9.7GB/Sec
(High Performance)
Database Restore Rate
Restoring 1TB TPCh data over IB connections
1 SQLIO Instance for each LUN in an HSA
Technical white paper
Page 8
Licensing and support considerations
The server used as the Data Integration Platform is not part of the HPE ConvergedSystem 300 for Microsoft Analytics Platform. It is essential to
understand that the licensing and support considerations for this server and any other ancillary servers are separate from that of the
ConvergedSystem.
Licensing considerations
The Data Integration Platform is a completely separate component from your HPE ConvergedSystem 300 for Microsoft Analytics Platform.
Therefore, you must purchase licenses for all software that will be used on this server.
The list of software that needs to be licensed will include:
• HPE iLO Advanced 1-Server including 1 yr. 24x7 software support
– For optimal management of the hardware, HPE iLO is strongly recommended
• Microsoft® Windows Server® 2012 R2 Standard Edition (x64 version)
– SQL Server 2014 (If SSIS is to be used)
You must either provide your own licenses or purchase these licenses from your authorized reseller (or directly from HPE if ordering a new
server).
Support considerations
The Data Integration Platform is not part of your HPE ConvergedSystem 300 for Microsoft Analytics Platform support contract with either HPE
or Microsoft. You must purchase support separately for all hardware and software components that will be used on the Data Integration Platform.
General considerations
It is highly recommended (though not required) that the Data Integration Platform is connected to the InfiniBand fabric for connectivity to the
ConvergedSystem. Using the high speed data interconnect will maximize the performance of the backup, restore, and data loading operations.
Any modifications to the Ethernet or InfiniBand switch configurations must only be to the ports that are being used by the Data Integration
Platform (e.g., setting VLANs) in the rack and not global switch configuration changes or changes to the ports used by the solution components.
Therefore, it is recommended that the Data Integration Platform is placed in the same rack or in an adjoining rack as the HPE ConvergedSystem
300 for Microsoft Analytics Platform. There is no requirement in terms of location, however if placed within the same rack, we highly recommend
that you order a separate PDU power source for the Data Integration Platform. If this is not possible, then there is a risk that any change in power
balancing, overloading of PDUs, etc. could impact the system and impact the power availability envelopes that have been designed in the rack.
Secondly, there is a self-imposed limitation to upgrading within the HPE ConvergedSystem 300 for Microsoft Analytics Platform rack that could
require removing the Data Integration Platform from the rack, if future expansion is desired or required.
Recommended hardware for the Data Integration Platform
HPE recommends the following hardware solution and configuration for a Data Integration Platform:
For a minimum configuration (1/4 rack which has 1 data scale unit):
• One (1) HPE ProLiant DL380 Gen9 (2x E5-2667 v4) Server including:
– 64GB RAM (Eight (8) HPE 8GB (1R x4) PC4-2400R-40 Memory)
– Twelve (12) HPE 1TB 6G SAS 7.2K 3.5in
– HPE IB FDR/EN 10/40Gb 2P 544+QSFP Adapter
– HPE P440ar/2GB FBWC 2P SAS Cntlr
– HPE SAS Expander
– 2x HPE IB FDR QSFP Copper Cables (to connect to the InfiniBand Switches)
– Appropriate Ethernet Networking Components (see below)
Technical white paper
Page 9
See Appendix A for complete information and part numbers.
If additional storage is required, you should select the HPE D6000 Disk Enclosure with the following options:
• For ½ rack with 2 data scale units or ¾ rack with 3 data scale units:
– Twenty-four (24) HPE 1TB 6G SAS 7.2K 3.5in
– 1x HPE Smart Array P841/4GB FBWC Controller
– 1x HPE D6000 Disk Enclosure
– 1x HPE D6000 Dual I/O Module Option Kit
– 4x HPE Ext Mini SAS HD 1m Cable
• For full rack with 3 data scale units:
– Forty-eight (48) HPE 1TB 6G SAS 7.2K 3.5in
OR
– Forty-eight (48) HPE 600GB 6G SAS 15K 3.5in (High Performance Option)
– 1x HPE Smart Array P841/4GB FBWC Controller
– 1x HPE D6000 Disk Enclosure
– 1x HPE D6000 Dual I/O Module Option Kit
– 4x HPE Ext Mini SAS HD 1m Cable
• For Configurations larger than a full rack (4+ scale units):
– Seventy (70) HPE 1TB 6G SAS 7.2K 3.5in
OR
– Seventy (70) HPE 600GB 6G SAS 15K 3.5in (High Performance Option)
– 1x HPE Smart Array P841/4GB FBWC Controller
– 1x HPE D6000 Disk Enclosure
– 1x HPE D6000 Dual I/O Module Option Kit
– 4x HPE Ext Mini SAS HD 1m Cable
If the storage space required exceeds the estimated disk storage needs of current forecasts, consider ordering 2, 3, 4 or 6TB disks instead. If
selecting larger drives, ensure that all the drives are of equal size to eliminate wasted drive space.
Recommended networking configurations for the Data Integration Platform
The default (and strongly recommended) configuration for connectivity to the HPE ConvergedSystem 300 for Microsoft Analytics Platform
solution is to connect directly to the InfiniBand switches for high data throughput for backup, and loading performance.
If connecting directly to the InfiniBand network is not an option, the secondary strategy would be to connect via the 10Gb Ethernet network
interface. Again, a direct connection to the switches is strongly recommended.
As a last (and not recommended) option, connectivity can be provided by connecting to the 1Gb management network of the system directly.
Table 3 below summarizes the connectivity options to the CS300 appliance.
Technical white paper
Page 10
Table 3. Connectivity overview
CONNECTIVITY TO THE HPE CONVERGEDSYSTEM 300 FOR
MICROSOFT ANALYTICS PLATFORM
InfiniBand Connection x 2 (one to each switch)
This is the recommended configuration
10Gb Ethernet Connection x 2 (one to each switch)
1Gb Ethernet Connection x 2 (one to each switch)
This is the least recommended configuration
Additionally, direct connectivity to the customer network over either 10Gb or 1Gb Ethernet from the Data Integration Platform is the
recommended configuration. This will provide the fastest data transfer, loading, and backup speeds.
The recommended InfiniBand connections (and cabling) are built into the parts list above. If the connectivity to the customer network is
connecting via 10Gb Ethernet, then you would add a 10Gb Ethernet network card into the order (and the necessary cables).
Note that the cable length for the cables will depend upon the location of the solution and distance to customer connectivity endpoints. Table 4
below depicts the connection options to the customer’s network.
Table 4. Customer network Connectivity
CONNECTIVITY TO THE CUSTOMER NETWORK
Redundant (or LACP) connectivity via 10Gb Ethernet
This is the recommended configuration
A single 10Gb connection
Redundant (or LACP) connectivity via 1Gb Ethernet
A single 1Gb connection is also an option
Network connectivity specifics
For the InfiniBand network, Figure 2 shows the current network port usage. Any of the ports in green (19-30) are available to connect to the
system. Ensure that you use IP addresses that are in the correct (customer-specified) ranges for each switch, and verify that the addresses are
not currently in use.
Figure 2. The InfiniBand switches cable wiring in an HPE ConvergedSystem 300 for Microsoft Analytics Platform (Rack #1)
Technical white paper
Page 11
For the Ethernet network, Figure 3 shows the current 1/10/40Gb network port usage. The green port (#23) is available to connect to the system
(1Gb Ethernet). Ensure that you use IP addresses that are in the correct (customer-specified) ranges for each switch, and verify that the
addresses are not currently in use.
Figure 3. The Ethernet switches cable wiring in an HPE ConvergedSystem 300 for Microsoft Analytics Platform (Rack #1)
For the 10Gb Ethernet network, Figure 4 shows the current Multi-Rack 10Gb connectivity. Again, the green “customer network” ports are
available for 10Gb connectivity, if any aren’t being used for rack connectivity to the customer network. Ensure that you use IP addresses that are
in the correct (customer-specified) ranges for each switch, and verify that the addresses are not currently in use.
Figure 4. 10Gb Ethernet Connectivity for the HPE ConvergedSystem 300 for Microsoft Analytics Platform
Technical white paper
Page 12
Recommended software for the Data Integration Platform
HPE recommends the following software for a Data Integration Platform:
• HPE iLO Advanced 1-Server including 1 yr. 24x7 software support
– For optimal management of the hardware, HPE iLO is strongly recommended
• Windows Server 2012 R2 Standard Edition (x64 version)
– Windows Server should be purchased/licensed as part of the purchase from HPE. However, this version of Microsoft Windows Server is
specifically requested because of the support for the InfiniBand firmware and drivers.
– Enable the Application Server and File and Storage Services Roles
– Enable Remote Desktop Services
• Version 4.85 or newer of the WinOF Driver for the HPE InfiniBand FDR/EN 10/40Gb Dual Port 544+QSFP Adapter
– 5.10 is the most current version (as of press time) for the 544+QSFP adapter.
– The driver is named: MLNX_VPI_WinOF-5_10_All_win2012R2_x64.exe
– Any version from version 4.85 forward can be used.
• DWLoader.exe
– This is the command-line data loading software that is included with your HPE ConvergedSystem 300 for Microsoft Analytics Platform
purchase. The software is on the system in the ClientTools.msi file.
• SQL Server Data Tools
– The SQL Server Data Tools are the primary SQL interface to the Parallel Data Warehouse software.
– Only the Visual Studio 2012 version of the SQL Server Data Tools is supported with Parallel Data Warehouse.
– The download is at: http://msdn.microsoft.com/en-us/jj650015
Note
The Data Integration Platform does not integrate into the domains on the HPE ConvergedSystem 300 for Microsoft Analytics Platform. The Data
Integration Platform should either belong to a customer domain or be a stand-alone workgroup server.
Caveats
The hardware configuration for the HPE ConvergedSystem 300 for Microsoft Analytics Platform is preconfigured at the factory, and any
topology changes need to be made by Microsoft or HPE-authorized personnel.
The customer may deploy the Data Integration Platform in the HPE ConvergedSystem 300 for Microsoft Analytics Platform rack provided it
follows the recommendations in this white paper.
• Configurations outside of the recommendations in this white paper may impact core services such as performance, security, and high
availability, and are not supported.
• The Data Integration Platform is not supported as part of the HPE ConvergedSystem 300 for Microsoft Analytics Platform, and needs to be
supported with separate support and condition. Any issues would not flow through the standard HPE/Microsoft collaborative support
model. HPE and Microsoft reserve the right to request that any additional equipment be shut down during troubleshooting if there is a
possible correlation or potential issue that is causing problems with the HPE ConvergedSystem 300 for Microsoft Analytics Platform
functionality.
• Any modifications to the Ethernet or IB switch configurations must only be to the ports that are being used by the additional equipment (e.g.,
setting VLANs) in the rack and not global switch configuration changes or changes to the ports used by appliance components.
Technical white paper
Page 13
Appendix A – Recommended configuration for a Data Integration Platform (with part numbers)
This appendix documents the specific parts list to order a “Data Integration Platform” for your HPE ConvergedSystem 300 for Microsoft
Analytics Platform solution.
Note
This configuration does not ship with any preconfigured software or with an OS installed. Any software or configuration installation is performed
on-site defined by the customer requirements. It is strongly recommended that you order the desired version of Windows Server with the
hardware order.
Note
Part numbers are at time of publication and subject to change. The bill of materials does not include complete support options or other rack and
power requirements. If you have questions regarding ordering, please consult with your HPE Reseller or HPE Sales Representative for more
details. hpe.com/us/en/services/consulting.html
Two BOMs are presented below. The first BOM is used when connecting the DIP server to a customer network through 1Gb Ethernet. The
second BOM is required when connecting through 10Gb Ethernet.
Table 5. DIP Server BOM (1GbE Connectivity)
HPE Data Integration Platform (1GbE Customer Connectivity)
QTY
PART NUMBER
DESCRIPTION
1
719061-B21
HPE ProLiant DL380 Gen9 CTO Chassis (12LFF)
1
817947-L21
Eight-Core Intel Xeon Processor Model E5-2667v4 (3.2 GHz, 25MB Level 3 Cache, 135W)
1
817947-B21
Additional E5-2667v4 CPU Kit
1
749974-B21
HPE Smart Array P440ar 12Gb 2-port FIO adptr
1
727250-B21
HPE 12Gb SAS Expander Card w/cables
1
719073-B21
3 Slot Expansion
1
665240-B21
HPE 366FLR 4 port FlexLom Adapter
1
764284-B21
HPE IB FDR 2-port 544+QSFP Adapter
8
805347-B21
8 GB Reg DDR4-2400R CAS 17 Kit
2
720479-B21
HPE 800W Flexslot Platinum HPE Power Supply
1
512485-B21
One (1) HPE iLO Adv 1-Svr incl 1yr TS&U SW
Technical white paper
Page 14
OR (For customers requiring 10GbE customer connectivity instead of 1GbE)
Table 6. DIP Server BOM (10GbE Connectivity)
HPE Data Integration Platform (10GbE Customer Connectivity)
QTY
PART NUMBER
DESCRIPTION
1
719061-B21
HPE ProLiant DL380 Gen9 CTO Chassis (12LFF)
1
817947-L21
Eight-Core Intel Xeon Processor Model E5-2667v4 (3.2 GHz, 25MB Level 3 Cache, 135W)
1
817947-B21
Additional E5-2667v4 CPU Kit
1
749974-B21
HPE Smart Array P440ar 12Gb 2-port FIO adptr
1
727250-B21
HPE 12Gb SAS Expander Card w/cables
1
719073-B21
3 Slot Expansion
1
727054-B21
HPE 562FLR-SFP+ 2 port FlexLom Adapter
1
764284-B21
HPE IB FDR 2-port 544+QSFP Adapter
8
805347-B21
8 GB Reg DDR4-2400R CAS 17 Kit
2
720479-B21
HPE 800W Flexslot Platinum HPE Power Supply
1
512485-B21
One (1) HPE iLO Adv 1-Svr incl 1yr TS&U SW
Drives
If ¼ rack (2 node) configuration: Twelve (12) HPE 6G SAS 7.2K 3.5in (Select 1, 2, 3, 4, or 6TB)
Table 7. Quarter Rack Internal Drives
QTY
PART NUMBER
DESCRIPTION
12
652753-B21
HPE 1TB 6G SAS 7.2K 3.5in SC MDL HDD
12
652757-B21
HPE 2TB 6G SAS 7.2K 3.5in SC MDL HDD
12
652766-B21
HPE 3TB 6G SAS 7.2K 3.5in SC MDL HDD
12
695510-B21
HPE 4TB 6G SAS 7.2K 3.5in SC MDL HDD
12
761477-B21
HPE 6TB 6G SAS 7.2K 3.5in SC MDL HDD
If the solution is larger than ¼ rack then add the following components Plus disks (either (a) or (b) below):
Table 8. External Storage Hardware
QTY
PART NUMBER
DESCRIPTION
1
726903-B21
HPE Smart Array P841/4GB FBWC Controller
1
QQ695A
HPE D6000 Disk Enclosure
1
QQ696A
HPE D6000 Dual I/O Module Option Kit
4
716189-B21
HPE Ext Mini SAS HD 1m Cable
Technical white paper
Page 15
a. If ½ rack (4 nodes) or ¾ rack (6 nodes) configuration: Twenty-four (24) HPE 6G SAS 7.2K 3.5in (Select 1, 2, 3, 4 or 6TB)
Table 9. External Drives (Half Rack to Three Quarter Rack)
QTY
PART NUMBER
DESCRIPTION
24
507614-B21
HPE 1TB 6G SAS 7.2K 3.5in SC MDL HDD
24
507616-B21
HPE 2TB 6G SAS 7.2K 3.5in SC MDL HDD
24
625031-B21
HPE 3TB 6G SAS 7.2K 3.5in SC MDL HDD
24
693689-B21
HPE 4TB 6G SAS 7.2K 3.5in SC MDL HDD
24
782669-B21
HPE 6TB 6G SAS 7.2K 3.5in SC MDL HDD
b. If full rack (8 nodes) configuration: Forty-eight (48) HPE 6G SAS 7.2K 3.5in (Select 1, 2, 3, 4 or 6TB)
Table 10. External Drives (Full Rack)
QTY
PART NUMBER
DESCRIPTION
48
507614-B21
HPE 1TB 6G SAS 7.2K 3.5in SC MDL HDD
48
507616-B21
HPE 2TB 6G SAS 7.2K 3.5in SC MDL HDD
48
625031-B21
HPE 3TB 6G SAS 7.2K 3.5in SC MDL HDD
48
693689-B21
HPE 4TB 6G SAS 7.2K 3.5in SC MDL HDD
48
782669-B21
HPE 6TB 6G SAS 7.2K 3.5in SC MDL HDD
OR
For the High Performance Storage Option: Forty-eight (48) HPE 6G SAS 15K 3.5in:
Table 11. External Drives (Full Rack – High Performance)
QTY
PART NUMBER
DESCRIPTION
48
516828-B21
HPE 600GB 6G SAS 15K 3.5in Ent HDD
c. If Larger than a full full rack (10+ node) configuration: Seventy (70) HPE 6G SAS 7.2K 3.5in (Select 1, 2, 3, 4 or 6TB)
Table 12. External Drives (Full Rack+)
QTY
PART NUMBER
DESCRIPTION
70
507614-B21
HPE 1TB 6G SAS 7.2K 3.5in SC MDL HDD
70
507616-B21
HPE 2TB 6G SAS 7.2K 3.5in SC MDL HDD
70
625031-B21
HPE 3TB 6G SAS 7.2K 3.5in SC MDL HDD
70
693689-B21
HPE 4TB 6G SAS 7.2K 3.5in SC MDL HDD
70
782669-B21
HPE 6TB 6G SAS 7.2K 3.5in SC MDL HDD
Technical white paper
Page 16
OR
For the High Performance Storage Option: Seventy (70) HPE 6G SAS 15K 3.5in:
Table 13. External Drives (Full Rack+ – High Performance)
QTY
PART NUMBER
DESCRIPTION
70
516828-B21
HPE 600GB 6G SAS 15K 3.5in Ent HDD
Please be sure and add the necessary cables for network connectivity to the InfiniBand and Ethernet switches in the solution, and select the
cables with the appropriate length. Use the following tables when selecting InfiniBand, 10Gb Ethernet and 1Gb Ethernet cables.
Table 14. InfiniBand Cables
InfiniBand cables
PART NUMBER
DESCRIPTION
670759-B21
HPE 0.5M IB FDR QSFP Copper Cable
670759-B22
HPE 1M IB FDR QSFP Copper Cable
670759-B23
HPE 1.5M IB FDR QSFP Copper Cable
670759-B24
HPE 2M IB FDR QSFP Copper Cable
670759-B25
HPE 3M IB FDR QSFP Copper Cable
670760-B22
HPE 5M IB FDR QSFP Optical Cable
670760-B23
HPE 7M IB FDR QSFP Optical Cable
Table 15. 10Gb Ethernet Cables
10Gb Ethernet cables
PART NUMBER
DESCRIPTION
JD096C
HPE X240 10G SFP+ SFP+ 1.2m DAC Cable
JD097C
HPE X240 10G SFP+ SFP+ 3m DAC Cable
JG081C
HPE X240 10G SFP+ SFP+ 5m DAC Cable
JC784C
HPE X240 10G SFP+ 7m DAC Cable
Table 16. 1Gb Ethernet Cables
1Gb Ethernet cables
PART NUMBER
DESCRIPTION
C7533A
HPE Ethernet 4ft CAT5e RJ45 M/M Cable
C7535A
HPE Ethernet 7ft CAT5e RJ45 M/M Cable
C7536A
HPE Ethernet 14ft CAT5e RJ45 M/M Cable
C7537A
HPE Ethernet 25ft CAT5e RJ45 M/M Cable
Technical white paper
Page 17
The transceivers must also be ordered for the connections to switches. You will need (at a minimum) two 10Gb Ethernet transceivers to connect
to the 10Gb ports on the HPE 5900 Ethernet Switches.
Supported transceivers include:
Table 17. 10Gb Transceivers
HPE 5900 10Gb Ethernet transceivers
PART NUMBER
DESCRIPTION
JG234A
HPE X130 10G SFP+ LC ER 40km Transceiver
JD092B
HPE X130 SFP+ LC SR Transceiver
JD093B
HPE X130 SFP+ LC LRM Transceiver
JD094B
HPE X130 SFP+ LC LR Transceiver
JD095C
HPE X240 10G SFP+ SFP+ 0.65m DAC Cable
JD096C
HPE X240 10G SFP+ SFP+ 1.2m DAC Cable
JD097C
HPE X240 10G SFP+ SFP+ 3m DAC Cable
JC784C
HPE X240 10G SFP+ 7m DAC Cable
JG081C
HPE X240 10G SFP+ SFP+ 5m DAC Cable
Technical white paper
Page 18
Resources and additional links
HPE ConvergedSystem 300 for Microsoft Analytics Platform:
hpe.com/products/cs300aps
HPE ConvergedSystem 300 for Microsoft Analytics Platform System:
Solution architecture and reference guide for the AU5 release:
http://h20195.www2.hpe.com/V2/GetDocument.aspx?docname=4AA5-9215ENW
To help us improve our documents, please provide feedback at hpe.com/contact/feedback.
Sign up for updates
Rate this document
© Copyright 2014-2016 Hewlett Packard Enterprise Development LP. The information contained herein is subject to change without
notice. The only warranties for HPE products and services are set forth in the express warranty statements accompanying such products
and services. Nothing herein should be construed as constituting an additional warranty. HPE shall not be liable for technical or editorial
errors or omissions contained herein.
Microsoft and Windows Server are trademarks of the Microsoft group of companies. Intel and Xeon are trademarks of Intel Corporation in
the U.S. and other countries.
4AA5-4660ENW, April 2016, Rev. 7
Download PDF
Similar pages