SPARC T4-4 Beats 8-CPU IBM POWER7 on TPC-H @3000GB Benchmark
- by Brian
Oracle's SPARC T4-4 server delivered a world record TPC-H @3000GB
benchmark result for systems with four processors.
This result beats eight processor results from IBM (POWER7) and
HP (x86). The SPARC T4-4 server also delivered better
performance per core than these eight processor systems from IBM
and HP.
Comparisons below are based upon system to system comparisons,
highlighting Oracle's complete software and hardware solution.
This database world record result used
Oracle's Sun Storage 2540-M2 arrays (rotating disk) connected to
a SPARC T4-4 server running Oracle Solaris 11 and Oracle Database
11g Release 2 demonstrating the power of
Oracle's integrated hardware and software solution.
The SPARC T4-4 server based configuration achieved a
TPC-H scale factor 3000 world record for four processor systems
of 205,792 QphH@3000GB with price/performance of $4.10/QphH@3000GB.
The SPARC T4-4 server with four SPARC T4 processors (total of 32 cores)
is 7% faster than the
IBM Power 780 server with eight POWER7 processors (total of 32 cores)
on the TPC-H @3000GB benchmark.
The SPARC T4-4 server is 36% better in price performance compared
to the IBM Power 780 server on the TPC-H @3000GB Benchmark.
The SPARC T4-4 server is 29% faster than the
IBM Power 780 for data loading.
The SPARC T4-4 server is up to 3.4 times faster than
the IBM Power 780 server for the Refresh Function.
The SPARC T4-4 server with four SPARC T4 processors
is 27% faster than the HP ProLiant DL980 G7 server
with eight x86 processors on the TPC-H @3000GB benchmark.
The SPARC T4-4 server is 52% faster than the HP ProLiant DL980 G7 server
for data loading.
The SPARC T4-4 server is up to 3.2 times faster than the
HP ProLiant DL980 G7 for the Refresh Function.
The SPARC T4-4 server achieved a peak IO rate
from the Oracle database of 17 GB/sec.
This rate was independent of the storage used, as demonstrated
by the TPC-H @3000TB benchmark
which used twelve Sun Storage 2540-M2 arrays (rotating disk)
and the TPC-H @1000TB benchmark which used
four Sun Storage F5100 Flash Array devices (flash storage).
[*]
The SPARC T4-4 server showed linear scaling from TPC-H @1000GB
to TPC-H @3000GB. This demonstrates that the SPARC T4-4 server
can handle the
increasingly larger databases required of DSS systems.
[*]
The SPARC T4-4 server benchmark results demonstrate a complete solution
of building Decision Support Systems including data loading, business
questions and refreshing data. Each phase usually has a time
constraint and the SPARC T4-4 server shows superior performance during
each phase.
[*] The TPC believes that comparisons of results published
with different scale factors are
misleading and discourages such comparisons.
Performance Landscape
The table lists the leading TPC-H @3000GB results for non-clustered
systems.
TPC-H @3000GB, Non-Clustered Systems
System Processor P/C/T – Memory
Composite(QphH)
$/perf($/QphH)
Power(QppH)
Throughput(QthH)
Database
Available
SPARC Enterprise M9000
3.0 GHz SPARC64 VII+
64/256/256 – 1024 GB
386,478.3
$18.19
316,835.8
471,428.6
Oracle 11g R2
09/22/11
SPARC T4-4
3.0 GHz SPARC T4
4/32/256 – 1024 GB
205,792.0
$4.10
190,325.1
222,515.9
Oracle 11g R2
05/31/12
SPARC Enterprise M9000
2.88 GHz SPARC64 VII
32/128/256 – 512 GB
198,907.5
$15.27
182,350.7
216,967.7
Oracle 11g R2
12/09/10
IBM Power 780
4.1 GHz POWER7
8/32/128 – 1024 GB
192,001.1
$6.37
210,368.4
175,237.4
Sybase 15.4
11/30/11
HP ProLiant DL980 G7
2.27 GHz Intel Xeon X7560
8/64/128 – 512 GB
162,601.7
$2.68
185,297.7
142,685.6
SQL Server 2008
10/13/10
P/C/T = Processors, Cores, Threads
QphH = the Composite Metric (bigger is better)
$/QphH = the Price/Performance metric in USD (smaller is better)
QppH = the Power Numerical Quantity
QthH = the Throughput Numerical Quantity
The following table lists data load times and refresh function
times during the power run.
TPC-H @3000GB, Non-Clustered Systems
Database Load & Database Refresh
System Processor
Data Loading(h:m:s)
T4Advan
RF1(sec)
T4Advan
RF2(sec)
T4Advan
SPARC T4-4
3.0 GHz SPARC T4
04:08:29
1.0x
67.1
1.0x
39.5
1.0x
IBM Power 780
4.1 GHz POWER7
05:51:50
1.5x
147.3
2.2x
133.2
3.4x
HP ProLiant DL980 G7
2.27 GHz Intel Xeon X7560
08:35:17
2.1x
173.0
2.6x
126.3
3.2x
Data Loading = database load time
RF1 = power test first refresh transaction
RF2 = power test second refresh transaction
T4 Advan = the ratio of time to T4 time
Complete benchmark results found at the TPC benchmark website
http://www.tpc.org.
Configuration Summary and Results
Hardware Configuration:
SPARC T4-4 server
4 x SPARC T4 3.0 GHz processors
(total of 32 cores, 128 threads)
1024 GB memory
8 x internal SAS (8 x 300 GB) disk drives
External Storage:
12 x Sun Storage 2540-M2 array storage, each with
12 x 15K RPM 300 GB drives, 2 controllers, 2 GB cache
Software Configuration:
Oracle Solaris 11 11/11
Oracle Database 11g Release 2 Enterprise Edition
Audited Results:
Database Size:
3000 GB (Scale Factor 3000)
TPC-H Composite:
205,792.0 QphH@3000GB
Price/performance:
$4.10/QphH@3000GB
Available:
05/31/2012
Total 3 year Cost:
$843,656
TPC-H Power:
190,325.1
TPC-H Throughput:
222,515.9
Database Load Time:
4:08:29
Benchmark Description
The TPC-H benchmark is a performance benchmark established by the
Transaction Processing Council (TPC) to demonstrate Data
Warehousing/Decision Support Systems (DSS). TPC-H measurements are
produced for customers to evaluate the performance of various DSS
systems. These queries and updates are executed against a standard
database under controlled conditions. Performance projections and
comparisons between different TPC-H Database sizes (100GB, 300GB,
1000GB, 3000GB, 10000GB, 30000GB and 100000GB) are not allowed by the
TPC.
TPC-H is a data warehousing-oriented, non-industry-specific
benchmark that consists of a large number of complex queries typical
of decision support applications. It also includes some insert and
delete activity that is intended to simulate loading and purging data
from a warehouse. TPC-H measures the combined performance of a
particular database manager on a specific computer system.
The main performance metric reported by TPC-H is called the TPC-H
Composite Query-per-Hour Performance Metric (QphH@SF, where SF is the
number of GB of raw data, referred to as the scale factor). QphH@SF
is intended to summarize the ability of the system to process queries
in both single and multiple user modes. The benchmark requires
reporting of price/performance, which is the ratio of the total HW/SW
cost plus 3 years maintenance to the QphH. A secondary metric is the
storage efficiency, which is the ratio of total configured disk space
in GB to the scale factor.
Key Points and Best Practices
Twelve Sun Storage 2540-M2 arrays were used for the benchmark.
Each Sun Storage 2540-M2 array contains 12 15K RPM
drives and is connected to a single
dual port 8Gb FC HBA using 2 ports.
Each Sun Storage 2540-M2 array showed
1.5 GB/sec for sequential read operations and
showed linear scaling, achieving 18 GB/sec
with twelve Sun Storage 2540-M2 arrays.
These were stand alone IO tests.
The peak IO rate measured from the Oracle database was 17 GB/sec.
Oracle Solaris 11 11/11 required very little system tuning.
Some vendors try to make the point that storage ratios are of customer
concern. However, storage ratio size has more to do
with disk layout and the
increasing capacities of disks –
so this is not an important metric in
which to compare systems.
The SPARC T4-4 server and Oracle Solaris efficiently managed
the system load of over one thousand Oracle Database parallel
processes.
Six Sun Storage 2540-M2 arrays were mirrored to another six
Sun Storage 2540-M2 arrays on which all of the Oracle database files
were placed. IO performance was high and balanced across all the arrays.
The TPC-H Refresh Function (RF) simulates periodical refresh portion of
Data Warehouse by adding new sales and deleting old sales data.
Parallel DML (parallel insert and delete in this case) and
database log performance are a key for this function
and the SPARC T4-4 server outperformed both the IBM POWER7 server
and HP ProLiant DL980 G7 server. (See the RF columns above.)
See Also
Transaction Processing Performance Council (TPC) Home Page
Ideas International Benchmark Page
SPARC T4-4 Server
oracle.com
OTN
Oracle Solaris
oracle.com
OTN
Oracle Database 11g Release 2 Enterprise Edition
oracle.com
OTN
Sun Storage 2540-M2 Array
oracle.com
OTN
Disclosure Statement
TPC-H, QphH, $/QphH are trademarks of
Transaction Processing Performance Council (TPC).
For more information, see www.tpc.org.
SPARC T4-4 205,792.0 QphH@3000GB, $4.10/QphH@3000GB, available 5/31/12,
4 processors, 32 cores, 256 threads;
IBM Power 780 QphH@3000GB, 192,001.1 QphH@3000GB, $6.37/QphH@3000GB,
available 11/30/11, 8 processors, 32 cores, 128 threads;
HP ProLiant DL980 G7 162,601.7 QphH@3000GB, $2.68/QphH@3000GB
available 10/13/10, 8 processors, 64 cores, 128 threads.
