UPDATED 3.25.16 In case you are new here, I’m a big fan of blade servers, however there is one situation where you’ll probably want to avoid them.
In the good ole days, one of the reasons people considered blade servers was they offered a reduction in power. However, with the introduction of faster CPUs and memory, this is no longer the case. In fact, if power savings is the primary goal for your server purchase my recommendation would be to stick with 1U servers. To support my recommendation, I decided to do a power comparison between blade servers, 1U servers and Dell’s new hyper-converged server (the Dell PowerEdge FX2.)
The base server in the power comparison was configured as follows:
- 2 x Intel E5-2660 v3 CPUs (2.6GHz,25M Cache,9.60GT/s QPI,Turbo,HT,10C/20T (105W) Max Mem 2133MHz)
- 16 x 16GB RDIMM, 2133MT/s, Dual Rank, x4 Data Width
- 2 x 300GB 15K RPM SAS 6Gbps 2.5in Hot-plug Hard Drive
- 2 x 10GbE NICs
- 2 x Single Port 8Gb Fibre HBAs
To keep the comparison as equal as possible, I used 10GbE Pass-Through modules and 8Gb Fibre Pass-Through modules on the blade chassis and a pair of 10GbE Pass-Through Modules on the Dell PowerEdge FX2 chassis. (As a side note, the I/O modules in the blade server chassis typically add just a few watts each so they wouldn’t add much to the comparison.) To do the comparisons, I used Dell’s power calculator (ESSA) and focused on using Dell equipment. This exercise is not intended to promote Dell’s servers, or say that Dell’s design is more power efficient but to show a power comparison of different server form factors. With that said, the results were not too shocking. The 1U server used less power in every case, however when compared in large quantity (16), it came with 382 watts of a fully loaded blade chassis – or 24 watts per server. The AMPs and BTU/Hr were also very close, as shown in the charts below. Although the Dell PowerEdge FX2 was higher in all three areas compared to the 1U server, it did offer savings versus the full blade chassis in smaller increments (4 servers.)
Although blade (and hyper-converged) servers show a higher power requirement, keep in mind there are enhanced settings like power capping that can be utilized to help insure the servers stay within a required power envelope. These server designs also provide additional non-power related benefits like consolidated management and potential I/O port savings but if power savings is what you’re looking for, then blade servers may not be ideal for your needs. The results shown in this post may not reflect every configuration (i.e. lower power CPUs or fewer memory DIMMs) so please use power configuration tools available to you before deciding on the blade form factor that will work best for your workloads.
UPDATED: Charts removed based upon Legal Notice from Dell Energy Smart Solution Advisor (ESSA) tool. Additionally, it’s been pointed out that power tools, like ESSA, are for worst case power scenarios. In real world numbers, it’s been seen that blade servers will pull around 5-7% less power than rack servers, on average. In summary, review both rack and blade server options carefully and decide what works best for your environment.
Kevin Houston is the founder and Editor-in-Chief of BladesMadeSimple.com. He has over 17 years of experience in the x86 server marketplace. Since 1997 Kevin has worked at several resellers in the Atlanta area, and has a vast array of competitive x86 server knowledge and certifications as well as an in-depth understanding of VMware and Citrix virtualization. Kevin works for Dell as a Server Sales Engineer covering the Global Enterprise market.