If you are looking at moving workloads to the cloud – you should stop and read this first.  Deploying an on-premises private cloud built on blade servers could save some major time and offer cost savings.  Of course, there are use cases where public cloud makes sense, and others where having an on-premise solution makes sense.  Here is one of those cases.

In a Principled Technologies’ (PT)  full report titled, “Run big data analytics on a powerful on-premises Dell EMC PowerEdge FX2 solution and save money over three years“, PT tested Hadoop and Apache Spark-based data analytics workloads against large sets of data on both an on-premise Dell EMC PowerEdge FX2 solution as well as Amazon Web Services (AWS) Elastic Compute Cloud (ECs) configuration.  Some of the findings about the on-premise Dell EMC PowerEdge FX2 solution vs the public cloud offering:

• cost savings of up to 42% over three years
• processed more data per second
• completed a big data workload over six minutes faster (24% faster)

Breakdown of the Estimated Cost Savings

Here is how the report calculated the cost savings.

For the AWS solution costs, PT used pricing for the US-East/US Standard (Virginia) datacenter in the AWS Simple Monthly Calculator to estimate the cost of the Amazon ECS instances and storage instances they used in testing.  Here is the breakdown:

Amazon EC2 Instances
# of instances: 4
usage: 100% utilized/month
type: Linux on M4.16xlarge
billing option: 3 year, all-upfront reserved

Amazon EBS volumes

volumes: 8
volume type: Provisioned IOPs SSD (Io1)
storage (GB): 600
IOPS: 4,500

volumes: 4
volume type: General Purpose SSD (gp2)
storage (GB): 30

Pricing Estimate – Amazon EC2 Service

Compute: $0.00
EBS Volumes (monthly fee): $612.00
EBS IOPS (monthly fee): $2,340.00
Reserved Instances (one-time fee): $126,480.00
Amazon EC2 Service price Sub-Total: $129,432.00

Pricing Estimate – Amazon Support

Support for all AWS Services (monthly fee): $295.20
Support for Reserved Instances (one-time fee): $8,076.40
AWS Support Service price Sub-Total: $8,371.60

Total of one-time fees: $134,556.40
Total of monthly fees: $3,247.20

Total estimated costs for 3 year Amazon AWS and EC2 solution = $251,455.60

 

The Dell EMC PowerEdge FX2 solution consisted of:

• 1 x Dell EMC PowerEdge FX2s chassis with 2 x PowerEdge FN I/O modules
• 4 x PowerEdge FC640 blade server nodes (built with 2 x Intel Xeon SP 5120 CPUs; 12 x 16GB RAM, 2 x 800GB SAS SSDs, Intel X710 Dual-Port 10Gb Network Daughter Card and 3 year ProSupport Plus Mission Critical 4Hr Onsite Service)
• Red Hat Enterprise Linux software (included subscription for 3 years for each of the servers; total of 12)
• Datacenter power and cooling cost (assumed a $10.63 KwH running 24/7 for three years; cooling was calculated at 70% of power)
• Datacenter space costs (calculated based on 24 square feet per rack at a cost of $270 per square foot)
• Server administration costs (calculated based on 1% of a 3 year total compensation of a Server Administrator II)

Hardware list price = $118,391.67
Software price = $15,588.00
Datacenter power and cooling = $1,118.57
Datacenter space = $3,262.59
Server administration = $3,262.59

Total estimated costs for 3 year Dell EMC Solution = $145,344.21 (42% less than Amazon AWS)

 

Better Performance and Faster Completion Times

Saving money is one thing, but performance is another crucial piece.  In the report, PT ran 4 HiBench tests on the workloads.  HiBench refers to Intel HiBench 7.1 – a big data benchmark suite for Hadoop.  The output of the tools can demonstrate a solution’s processing speed, throughput, bandwidth, CPU utilization, data access patterns and other big data workload metrics.  From the results of the tests, the biggest margin of difference was with the Kmeans throughput analysis.  The Dell EMC solution delivered a throughput of more than 1,200 MB per second – a 36% throughput increase over Amazon AWS.

In the Latent Dirichlet allocation (LDA) benchmark, the on-premise solution using a Dell EMC PowerEdge FX2 solution completed the LDA workload in 18.9 minutes where the Amazon AWS solution took 25.1 minutes – a difference of six minutes.  To be fair, here are the time comparisons of all four benchmarks (lower time is better):

Bayes
Dell EMC – 2.8 minutes
AWS – 3.2 minutes

Kmeans
Dell EMC – 3.1 minutes
AWS – 4.3 minutes

LDA
Dell EMC – 18.9 minutes
AWS – 25.1 minutes

RandomForest
Dell EMC – 8.0 minutes
AWS – 8.9 minutes

Additional Reasons to Consider On-Premise Solutions

Principled Technologies’ also published a report titled, “Get the advantages of an on-premises cloud and use familiar tools”  where they summarize additional advantages of running an on-premise solution versus a public cloud.  In the report, they state in a public cloud you are locking yourself into a monthly or yearly contract.  Like a car lease, these contracts could have penalties for exceeding allotments which cost your organization more money.  In contrast, with an on-premise private cloud blade server solution, you pay up front and get a flexible server platform that can be sliced up and allocated based on needs and easily configured as needs change.  Additionally, on-premise solutions are more secure and will allow you to know exactly where your data is.  On-premise solutions also allow you to customize based on your needs versus a “one-size-fits-all public cloud option.

Conclusion

On a final note, the reports above are commissioned by Dell EMC (and I work for Dell EMC as well) but I hope it is enough information to empower you to take the time to compare costs savings of any public cloud solution with an on-premise solution be it Dell EMC, or any other vendor.  I’d love to know if you agree with Principled Technologies’ findings, so please leave comments below.

 

Kevin Houston is the founder and Editor-in-Chief of BladesMadeSimple.com.  He has 20 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 has worked at Dell EMC since August 2011 working as a Server Sales Engineer covering the Global Enterprise market from 2011 to 2017 and currently works as a Chief Technical Server Architect supporting the Central Region.