A shot in the ARM: how to improve application performance on cloud

A shot in the ARM: how to improve application performance on cloud

Posted by Geoff Davies on Jun 30, 2021 2:30:00 PM

ARM-based server offerings from AWS and Oracle offer price performance gains of up to 40%, here's how to identify suitable workloads.

In the early 1980s, with the personal computer market hotting up, IBM introduced the PC and Apple their "Lisa", both aimed at business users.

At the same time, a small company in England, called Acorn, assessed how to enter this lucrative business market. Acorn had produced the successful BBC Micro for home and educational users but it lacked the performance to compete with the PC and Lisa.

The team, led by computer scientist Sophie Wilson, set out with the ambition to produce a machine with 10 times the performance of the BBC Micro at the same price point. But, after analysing the CPU designs available, the engineers quickly established that performance came at a price.

A breakthrough came by combining research on Reduced Instruction Set Computers (RISC) from the University of California, Berkeley, with some of the successful design features of the BBC Micro and 6502 processor. In doing so the team produced the world's first practical RISC processor. On 26th April 1985 the Acorn RISC Machines - or ARM - processor was born.

Providing high performance at a low price point and power consumption also made ARM highly suitable for small devices. Thus by 2005, through extensive licensing deals, 98% of mobile phones ran on ARM processors.

These chips also became available in servers for specialised applications. Whereas traditional x86 processors address performance by scaling up processor speed and capabilities, ARM-based servers scale-out by spreading the load across many smaller, lower power CPUs.

In the traditional data centre world, building out an ARM-based capability requires detailed analysis and a strong business case to justify the investment. But in the world of cloud, ARM-based servers are readily available on-demand in the shape of AWS's Graviton2 and Oracle Cloud Infrastructure (OCI) Ampere servers.

Whilst ARM processors are available for everyone they're not suitable for everything. For instance, you can't deploy Microsoft Windows Server onto them, instead you need to run a modern version of Linux or BSD as the OS.  Moreover, changes need to be made to the software to run on ARM;  therefore you need access to the source code or applications should be built on suitable open-source components.

New call-to-action

The upside is that applications modified to run on Graviton2 or Oracle's Ampere will benefit from a significant cost/performance improvements. Indeed, AWS predict a staggering 40% better price performance over similar x86 based instances.

For individual applications it's easy to get started migrating to ARM, indeed AWS have recently launched a Graviton2 challenge that will enable you to get an application running on a Graviton2 instance in just 4 days (Note: competition closes August, 31st 2021 and is only open to US residents).

However, within a large enterprise application portfolio it can be challenging to identify suitable candidates, this is where AppScore can help. Being a data-based platform which enables deep insights into your application portfolio you can quickly and simply find where these candidates are.

Here's how to use filters in AppScore to identify a high-level set of suitable applications:

Then, dive deeper into this scope to determine a set of initial candidates based on:

  • Business criticality: business critical applications may not make good initial candidates, gain experience on less critical apps first.
  • Roadmap status (planned usage): choose applications that are strategic and have a mid- to long-term lifespan.
  • Age of application/update recency: the newer the code, or if it's under active maintenance, the easier it will be to update to run on ARM.
  • Size of server footprint: small server footprints may not present a good effort vs. cost benefit opportunity.

With your initial candidates selected, engage the application teams to start work on migrating to Ampere on Oracle, or Graviton2 on AWS and unlock those significant cost savings.

It really is as easy as that to quickly find ARM opportunities within your application portfolio.

For a demo on how to use AppScore to transform your apps portfolio for the cloud book a session here.