White Papers

37 BP1062 | Best Practices for Decision Support Systems with Microsoft SQL Server 2012 using Dell EqualLogic PS Series
Storage Arrays
one array. This was because there was no application cache or I/O writes involved with the IOMeter tool.
The Benchmark Factory TPC-H database test had SQL Server buffer cache and a small percentage (4%) of
I/O writes (Tempdb and Log files) involved causing the difference in throughput. In addition, the workloads
were not 100% sequential when the actual TPC-H database was accessed by multiple users.
6.4.2 Scaling only the arrays with constant user load – TPC-H
For the scalability tests, the eight partition configuration (shown in the table partition studies in section
6.2.2) was used because it provided a better execution time. Six user queries were run against a single
database deployed in single array (Test Configuration #1A, Figure 16) and two arrays (Test Configuration
#1B, Figure 16). The volume configuration shown in Figure 16 and the configuration parameters shown in
Table 8 were used for this test.
Volume layout for the SAN scaling studies Figure 16
Test parameters: SAN scaling Table 8
Configuration Parameters
EqualLogic SAN
Two PS6110X (2.5", 24 10 K SAS drives,900 GB)
RAID type
RAID 50
Volume Configurations
Test Configuration #1A
(Six users, one database, one
array)
•
One PS6110X (in one EqualLogic storage pool)
• 6 users (132 queries) from Benchmark Factory
• Database volumes: Refer to Figure 16
Test Configuration #1B
• Two PS6110X (in one EqualLogic storage pool)