WP_Data Management Best Practices

Unfortunately, TPC-H has received multiple criticisms, from both MPP database vendors and analysts, and it is

disappointing that some of the most popular vendors (Teradata, IBM Netezza, HP Vertica, Pivotal Greenplum) have

basically shunned it. Only a handful of DBMS vendors continue measuring their systems, among which there are

Microsoft (although not for their MPP database), Oracle, Actian, Exasol, and Sybase IQ. Nevertheless, MPP

databases (“Clustered systems” in the TPC-H terminology) have beenmeasured to be generally superior to traditional

non-clustered systems, both in performance and in price/performance. Here is a snapshot of the results for the 3000

GB size:

You can see that the ‘cluster’ column of the first product is a ‘Y’ and the two others are ‘N’. Details on each system can

be obtained by clicking on the link under the System column. For instance, the Exasol database results were run on a

Dell cluster of 28 bi-processor servers. The second line results, for the Actian database, were obtained on an HPE

SMP machine with 4 multi-core processors. Here are the results for the 10 TB (10,000 GB) database size, in where

MSFT SQLServer recently posted a world record in this TPC-H 10 TB “non-clustered” category . The ratio in terms of

QphHwidens fromabout 8:2 to 10:1 with the size increase, an expected result.

—

Hardware infrastructure

. Assessing the server and network infrastructure required to support the desired

performance and scalability is addressed in

pages 362 and 363. It is best to keep a factor of safety margin

[1]

while planning for hardware by considering factors like back-up storage, multi-user concurrent load,

unplanned shut-downs, network issues etc. Hard-disk, CPU cores and memory are predominant factors while

arriving at optimal hardware needs.

W H I T E P A P E R

This was on a Lenovo x3850 X6 system that was configured with four Intel Xeon E7-8890 v4 processors at 2.20 GHz (4 processors/96

cores/192 threads), 6 TB of memory, and eight SanDisk ioMemory SX350 6.4TB PCIe flash adapters.

3,000 GB Results

Rank Company

System

QphH Price/QphH Watts/KQphH System

Availability

Database

Operating System

Date

Submitted Cluster

Dell PowerEdge R720xd

using EXASolution 5.0

7,808,386

.15 USD

09/24/14 EXASOL EXASolution

5.0

EXASOL EXA Cluster OS

5.0

09/23/14

HPE ProLiant DL580

Gen9 Actian Vector 5.0

2,140,307

.38 USD

07/31/16 Actian Vector 5.0

Red Hat Enterprise Linux

Server 7.2

06/02/16

Cisco UCS C460 M4

Server

1,071,018

.60 USD

06/01/16 Microsoft SQL Server

2016 Enterprise Edition

Microsoft Windows Server

2012 R2 Standard Edition 05/14/16

www.persistent.com

10,000 GB Results