In my last blog post, we talked about the importance of standard benchmark testing and the role it can play in helping customers in their purchase process. HP's benchmark page references their Proliant DL180 as besting the PowerEdge 2950 III by 14%. Naturally we took this to heart and decided to do some investigating.
According to HP's website, the Proliant DL180 is positioned as ideal for small to medium business. The DL180 is limited to 16 GB of memory, redundant non-hot pluggable power supplies, non-redundant fans, and one integrated NIC. This puzzles me as to why they would compare this entry level server to the PE 2950 III with 32GB of memory, redundant hot-pluggable power supplies, redundant fans, and two integrated NICs. Another point that popped out to me was why HP would use 5.4K RPM SATA hard drives that only produce 65 IOPS vs. the Dell enterprise class 10K RPM SAS hard drives that produce 190 IOPS. In addition, when using the SATA drive, the data integrity is only verified on the drive's writes while SAS drives verifies on both writes and reads.
If a customer is looking to use servers for enterprise business critical applications, I really doubt they would opt for lowering performing SATA drives versus SAS drives. Bottom-line, enterprise applications require higher performance drives, so HP's offering is not balanced. HP is lowering power by using lowest-rated parts regardless of the loss of performance. This is detrimental to most I/O intensive applications.
This clearly reinforces the need to have real world configurations for benchmarks instead of trying to move the needle by using unrealistic configurations. We as systems providers owe it to customers who use these 3rd party testing services as a key element in measuring which solution will work best in their environment. We need to provide "apples to apples" comparisons when comparing our server product lines-real-world configurations that are enterprise ready.
I would ask customers, what do you think? And HP, how about a more realistic comparison of power efficiency results for the Proliant DL380 against the PowerEdge 2950 III.