Bare metal cloud IaaS
Bigstep has announced the findings of new benchmarking studies that
reveal how users can get up to 60 per cent more performance from any
infrastructure.
The studies sought to assess areas of potential performance improvement in infrastructure and used the
Linpack,
SysBench and
TPC-DS
standard hardware benchmarking tools to do so. The findings can be
applied to any infrastructure and Bigstep commercial director Ioana
Hreninciuc believes many IT directors are not seeing the best
performance from their infrastructure:
“We operate tens of thousands of servers and constantly test
performance, so we know that an infrastructure is the heartbeat of any
IT environment,” she said. “Some findings from the benchmarking are
hugely surprising – who would’ve thought that deactivating one of two
CPUs would increase performance by up to 15 per cent? Many users are not
seeing the performance they could and, with just a few tweaks, any
server could perform by up to 60 per cent better, avoiding costly
upgrades or migrations.”
The benchmarking addressed four main areas:
1. Deactivating one of two CPUs can improve database performance by up to 15 per cent
In benchmarks performed on several databases, as well as in SysBench and
TPC-DS tests, Bigstep found that single processor bare metal instances
can generate better performance than dual processor machines, all other
things being equal. This means that adding a second processor to a
machine can actually lower a database’s performance. According to
Hreninciuc, this is caused by the two processors using each other’s
memory, which takes longer as opposed to each CPU accessing only its
own: “A way to counteract this limitation of existing hardware
infrastructure is to configure the OS to use only Local Memory Access
for each processor.”
2. OS choice can improve the performance of applications by up to 20 per cent
In its Linpack benchmarks, Bigstep found that CentOS 6.4 can yield up to
16-20 per cent better performance, when paired against Ubuntu 12 LTS.
The tests were done using default options and kernel settings. “By
checking the differences between kernel versions and settings and
configuring the OS to match those of the better performing system, users
can achieve a significant improvement in performance, without any
upgrade to the underlying hardware,” said Hreninciuc.
3. Improving memory frequency can increase performance by more than 20 per cent
Memory frequency is often overlooked as RAM is generally expected to be
‘fast enough’. However, the impact memory can have on computing
performance is underestimated. In Linpack benchmarks, Bigstep found that
replacing 1333 MHz DIMMs with 1866 MHz DIMMs from the same provider
increased overall computing performance (total no. of GigaFLOPs) by 20
per cent.
Applications that are heavily impacted by memory access time, such as
NoSQL databases, can be greatly affected by this metric. “If possible,
memory frequency should always be checked – as two bare metal machines
with otherwise identical specs, can register differences in performance
of up to 40 per cent, solely based on this metric,” advised Hreninciuc.
The physical placement of memory in slots can also greatly impact
performance. With the wrong placement, memory chips will not make use of
their maximum frequency and could even function at half their intended
performance. Some vendors provide online configurators for optimum
memory placement inside the bare metal server. Just by changing the
distribution of memory across their machines, users could achieve a
significant increase in frequency and performance.
4. Hyper-Threading can decrease performance for CPU intensive applications
Intel’s Hyper-Threading technology is meant to accelerate CPU
performance and in most cases it does that very well. However, in
applications that are very CPU intensive, deactivating Hyper-Threading
can yield 5-10 per cent better performance. In bare metal environments,
deactivating Intel Virtualization Technology (Intel VT-x) can also yield
an extra performance boost of up to 5 per cent.
“Additional performance is always valuable and applying these tips based
on our benchmarking could help any IT director obtain up to 60 per cent
more performance from their existing infrastructure,” concluded
Hreninciuc.
Bigstep’s product manager Alex Bordei will be discussing the benchmarking in more detail during an exclusive
O’Reilly webcast ]‘Getting the Most Out of Your NoSQL DB - Best Practices for Optimizing Infrastructure Performance and Budget’|
http://www.oreilly.com/pub/e/3143] on 7 August.
The webcast is hosted by O’Reilly’s Chief Data Scientist and Director of
Content Strategy for Data, Ben Lorica, and anyone interested in
attending can register for free
here.