Rackspace Cloud Said Performance & Meant It

To see the latest results, please click below.


See More Provider Comparisons Free


Benchmarking the Rackspace Evolution

Cloud Spectator ran system, vCPU and RAM benchmark tests on the new Rackspace Performance offering as well as the Next Generation and First Generation offerings. All server sizes were benchmarked to examine performance differences over a comprehensive data set that allowed analysis of different generations of offerings and different server sizes within the same offering. To view all results, please download the Full Report. Key findings and considerations are listed below.


Download the Full Report

View the Rackspace Server Offerings

Introduction

Rackspace has been an interesting provider to watch in regard to the evolution of its public cloud offering. We saw the First Generation cloud servers transition into one of the first OpenStack public cloud offerings in August of 2012. In early November 2013, Rackspace announced the release of the next phase in its public cloud – Performance Cloud Servers. Performance Cloud Servers will gradually roll out and replace Standard Cloud Servers as the default offering. Throughout this blog and the Full Report, Rackspace’s first public cloud offering will be referred to as First Generation, Rackspace’s original OpenStack offering will be referred to as Next Generation, and the Performance Cloud Servers will be referred to by name.

Performance Cloud Servers deliver more of what you would think: performance. Rackspace has updated its hardware and increased the capacity of its new virtual machines.
• RAID 10 Solid-state drives (SSDs)
• Intel Xeon E5 processors
• Machines with up to 10Gbps throughput
• Error-correcting code (ECC) memory
• Up to 32 vCPUs with 120GB of RAM

Whereas in their Next Generation offering, Rackspace’s largest server size was 30GB of RAM with 8vCPUs, Rackspace’s largest server now boasts 120GB of RAM with 32 vCPUs. Server sizes fall under two categories: “Performance 1” and Performance 2”. Performance 1 offerings range from 1 GB to 8 GB servers while Performance 2 servers range from 15 GB to 120 GB servers.

Other offering changes from Next Generation to Performance Cloud:
• Prices have been reduced (by one-third) for the 1GB through the 15GB offerings.
• The 4GB and 8GB servers have more vCPUs allocated to them, even with the price reduction.
• The 15 GB server has two less vCPUs allocated to it.
• The 512MB instance has been eliminated.

Key Findings

Performance Servers CPU scaling

Users running CPU-intensive workloads may consider the 4GB Performance 1 offering over the 15GB Performance 2, or the 8GB Performance 1 over the 30GB Performance 2. This is due to the vCPU allocation, which is equivalent between the 4GB and 15GB, and the 8GB and 30GB. Tests show similar CPU performance in those sets. Users can save up to $9,000 per year with this decision.

While the number of cores remain the same, memory scales by almost 4x, and price increases 4.25x. For CPU-intensive applications such as video encoding, results from the test (which can be found in the Full Report)found no difference in performance when the number of virtual cores remained the same; those types of applications may find more savings in the smaller, 4GB Performance Server option. The Video Encoding test did score higher on the 8 vCPU servers.


vCPUs Memory Cost per Hour
4 4GB $0.16
4 15GB $0.68
While the number of cores remain the same, memory scales by almost 4x, and price increases 4.25x. For CPU-intensive applications such as video encoding, results from the test (which can be found in the Full Report)found no difference in performance when the number of cores remained the same; those types of applications may find more savings in the smaller, 8GB Performance Server option. The Video Encoding test scored higher on these 8 vCPU servers than the 4 vCPU ones.


vCPUs Memory Cost per Hour
8 8GB $0.32
8 30GB $1.36
Single Core Performance

A Performance Server’s core outperforms both a First Generation and Next Generation core by 60% (with the exception of the 2GB Next Generation server – see “To Consider”).

Geekbench 3 Single Core Test Scores

Performance Score – 2169
Performance Score - 2169 — 100%

Next Gen Score – 1327
First Gen Score - 1327 — 61%

First Gen Score – 1379
Next Gen Score - 1327 — 64%

Data Center Performance Stability

The Next Generation servers were provisioned on different physical machines; the more modern 2GB RAM dual core scored 25% higher in CPU tests than the less modern dual core 4GB RAM offering. This may not be the case for the new Performance offerings, which all seem to run on Intel E5s within the North Virginia data center for more predictable VM performance.

Offering Processor Type Datacenter
2GB AMD Opteron 4332 HE (3.00 GHz) Dallas Fort Worth (DFW)
4GB AMD Opteron 4170 HE (2.10 GHz) Dallas Fort Worth (DFW)
Performance Servers RAM Bandwidth

Memory bandwidth of the Performance Servers increase by around 2.5x compared to the memory speeds of the Next Generation servers, creating a more suitable environment for databases running from memory.

RAMspeed/SMP Multi-Core Memory Performance

Performance Servers – 18442 MB/s
Performance Score - 2169 — 100%

Next Gen Servers – 7420 MB/s
First Gen Score - 1327 — 40%

First Gen Servers – 7040 MB/s
Next Gen Score - 1327 — 38%

To Consider

The Rackspace Next Generation 2GB virtual machine was provisioned on an AMD Opteron 4332 HE, a more modern processor than other Next Generation machines. This explains the increased single-core and multi-core performance of the 2GB machine.


While the same general pattern of performance emerges from each generation, the Unixbench test suite gives slightly different results from the Geekbench test suite because the Unixbench test suite considers performance of disk with its File Copy tests.

To view breakdowns of the Rackspace results from Geekbench tests, please visit the links in the Appendix of the Full Report.

This experiment was conducted during the initial release of the Rackspace Performance offering. Thus, the benchmark scores and measurements may not be reflective of fully user-saturated physical machines.

While the offerings are tiered in accordance to RAM amount, vCPUs vary among generations of the same-tiered offering. Costs remain similar, despite the difference in vCPU allocation. More information on server sizes can be found in the Appendix of the Full Report.

Download the Full Report