One of the most amazing features of the SQL Server Network Workload Tool (SSNWT) is to get execution results from different SQL Server environments to compare them, regardless any technical detail. In this way, SSNWT make easir to compare performance between different SQL Server instances, even having phisical or virtual servers, differents CPUs or disk subsystems, and so on.
In fact, one of the main reasons why I decided to develop the SSNWT was to compare performance running local vs from a virtual machine on the network, to identify any network bottleneck.
Therefore, I'll stop writting and post the first SSNWT execution results, so here you are!
Inserting 100.000 rows having a payload of 1000 bytes
Test 1 - GuilleSQL’s laptop
- Hardware Summary: HP Pavilion DV6 (CPU Intel i7-2630QM having 8 cores, 8GB RAM DDR3, 5400rpm SATAII disk)
- Software Summay: SQL Server 2008 R2 running on Windows Server 2008 R2
- Additional info: The test has been run on local, therefore, using Shared Memory protocol.
| Protocol |
Test Mode |
Threads |
Rows per Insert |
Partitioning |
Lock Escalation |
Row Locks |
Page Locks |
Seconds |
| Auto |
Secuential |
N/A |
N/A |
OFF |
N/A |
ON |
ON |
46 |
| Auto |
Multi-Threads |
8 |
N/A |
OFF |
N/A |
ON |
ON |
13,42 |
| Auto |
Multi-Threads |
10 |
1000 |
OFF |
N/A |
ON |
ON |
9,43 |
| Auto |
Multi-Threads |
10 |
1000 |
ON |
Auto |
OFF |
OFF |
7,47 |
Test 2 - Corporate SQL Server Cluster on a DataCenter located in Spain
- Hardware Summary: HP Blade (CPU 4 x AMD Opteron 6174 having 48 cores, 72GB RAM, HP XP shared Storage Subsystem connected by Fibel Channel)
- Software Summay: SQL Server 2008 R2 running on Windows Server 2008 R2
- Additional info: Because it’s a named instance running on Cluster, when selecting AUTO for the protocol, the TCP protocol is selected by default. So in this test, Shared Memory is not used, even running the test on local.
Running on local
| Protocol |
Test Mode |
Threads |
Rows per Insert |
Partitioning |
Lock Escalation |
Row Locks |
Page Locks |
Seconds |
| Auto |
Secuential |
N/A |
N/A |
OFF |
N/A |
ON |
ON |
71,74 |
| Auto |
Multi-Threads |
10 |
N/A |
OFF |
N/A |
ON |
ON |
12,66 |
| Auto |
Multi-Threads |
10 |
1000 |
ON |
Auto |
OFF |
OFF |
1,81 |
Running from a physical server on the network (in the same subnet)
| Protocol |
Test Mode |
Threads |
Rows per Insert |
Partitioning |
Lock Escalation |
Row Locks |
Page Locks |
Seconds |
| Auto |
Secuential |
N/A |
N/A |
OFF |
N/A |
ON |
ON |
91,21 |
| Auto |
Multi-Threads |
10 |
N/A |
OFF |
N/A |
ON |
ON |
15,55 |
| Auto |
Multi-Threads |
10 |
1000 |
ON |
Auto |
OFF |
OFF |
2,30 |
Running from a VMWare Virual Machine on the network (in the same subnet)
| Protocol |
Test Mode |
Threads |
Rows per Insert |
Partitioning |
Lock Escalation |
Row Locks |
Page Locks |
Seconds |
| Auto |
Secuential |
N/A |
N/A |
OFF |
N/A |
ON |
ON |
111,99 |
| Auto |
Multi-Threads |
10 |
N/A |
OFF |
N/A |
ON |
ON |
18,22 |
| Auto |
Multi-Threads |
10 |
1000 |
ON |
Auto |
OFF |
OFF |
2,51 |
As I said before, if you are interested, please send me feedback about your performace result and test environment.
That's all for today. I hope you enjoy the reading ! Cheers
| 
