To Malaysia-based Supahands, a workforce crowd-sourcing platform, cloud infrastructure services have freed up its time to focus on core activities and innovation projects, rather than worry about running out of computing and storage resources during peak periods.
For some years now, the startup company, which helps companies recruit remote workers to perform tasks such as training keyword-tagging algorithms, has been using load-balancing and auto-scaling services from cloud supplier Amazon Web Services (AWS) to keep pace with fluctuating demands from users.
The agility comes with the added benefit of keeping infrastructure costs in check, enabling companies such as Supahands to operate without having to grapple with the exorbitant fixed cost of setting up an on-premise system.
“To put it simply, if we were to compare the costs of on-premise versus cloud, we have saved around 95% in IT investment. Without a doubt, the cloud has made our business viable,” Supahands’ head of sales Greg Meehan told Computer Weekly.
Across the Asia-Pacific (APAC) region, more companies, including bigwigs like Southeast Asia’s largest bank DBS, are turning to cloud infrastructure services to augment or even replace their entire IT infrastructure in some cases.
According to IDC, infrastructure as a service (IaaS) will be the single largest category of public cloud spending in 2018, accounting for 47.6% of the overall cloud expenditure in APAC, with server spending slightly ahead of storage spending.
The huge market potential has led major cloud suppliers Amazon, Microsoft, Google, and more recently Alibaba, to expand their footprint in the region, whether it is opening more datacentres and availability zones, or speeding up the time it takes to bring new services to this part of the world.
Earlier this year, AWS and Google opened their third availability zones in Singapore, while Alibaba set up a new datacentre in Indonesia, Southeast Asia’s largest economy, as part of its global expansion plans.
While having multiple cloud suppliers eyeing a slice of the pie is good news for enterprises, choosing the right IaaS service is not always a straightforward affair, particularly as most offerings are largely similar.
“They all have similar capabilities, in terms of reliability and speed,” says Tim Sheedy, principal advisor of Ecosystm, a technology research firm. “For example, in terms of storage options, they all have more expensive storage for media access and deeply discounted versions for archived data.”
The highly competitive market also means all suppliers will try to match one another in features and price points. Sheedy notes that AWS and Google, for one, are looking at letting businesses perform analytics on data stored on cheaper storage, and that it will be a matter of weeks before others follow suit.
Start with a review
Before getting into cloud infrastructure services, organisations should first determine what exactly they want to achieve by the move.
“Public IaaS is not a magical solution for all things,” says Vish Iyer, vice-president for architectures at Cisco Asia-Pacific.
Vish Iyer, Cisco
“For it to be effective, it must be used as part of a larger digital and cloud strategy. They must first look at their business strategy, the digital strategy required to deliver it, their current and future data classifications, workload, applications and how they are built.”
The outcome of the internal review, Iyer adds, will determine the requirements for key IaaS features such as mobility, security, visibility, automation, application modernisation and portability, and regulatory and geographic requirements. “These decisions should be made with regards to the whole cloud/digital environment,” says Iyer.
More importantly, enterprises have to optimise their IT environments, right-size their workloads, enhance their security, network and technology frameworks, as well as align people, processes and toolsets.
Patrik Bihammar, cloud and enterprise lead at Microsoft Singapore, points out a number of factors that organisations should consider when evaluating cloud infrastructure offerings:
- What is the scale and geographic reach of the cloud service provider? Are they able to serve your business needs in all markets where you operate?
- Does your cloud provider have the enterprise experience and credibility you would expect, and would place your trust in them?
- What is the breadth and openness of the IaaS offerings?
- Does the cloud supplier have strong and strategic partnerships with industry leaders and players to extend the capabilities and applications available?
- As most organisations are not planning to move all their workloads to the public cloud, does the cloud supplier give you the flexibility to run IT in a hybrid model with some services on-premise and some on the public cloud?
- Does the cloud provider offer services and systems that enable efficiency and productivity within the workforce?
Stressing the importance of having a broad range of IaaS services, AWS’s head of solutions architecture for Malaysia, Philippines and partners, Santanu Dutt, says a cloud supplier should provide different options for varied use cases.
“A customer could have a standard use case to begin with, but six months later, it might need to run SAP applications and high-performance computing workloads, so it will need variations of CPU-memory instances, as well as gigabit networking and high memory instances,” Dutt says.
In terms of storage and databases, Dutt notes that cloud suppliers should offer solid-state drives and support for multiple databases: “Are they running various databases on the public cloud from a licensing or certification perspective? And can the cloud provider host the databases in full capacity?”
Just as important is the frequency of releases since most enterprises will have a longer-term roadmap that projects future technology requirements. “In 2017 alone, we had 1,430 public-facing services and feature releases, and any technology that customers ask for will be added over time as part of our agile development methodology,” Dutt says.
Related to the geographic reach – which must be wide enough to support an enterprise’s global footprint – is networking. Tim Synan, head of Google Cloud in Southeast Asia, says the search giant runs a high-quality private network with more than 100 global network points of presence.
“Google Cloud Platform also uses software-defined networking and distributed systems technologies to host and deliver services around the world,” says Synan. “When every millisecond of latency counts, Google ensures that content is delivered with the highest throughput, thanks to innovations like BBR congestion control intelligence.”
BBR, which stands for Bottleneck Bandwidth and Round-trip propagation time, is Google’s congestion control algorithm that is used to decide how fast to send data. Google says BBR has yielded 4% higher network throughput for its YouTube service, because the algorithm more effectively discovers and utilises bandwidth offered by the network.
Tim Sheedy, Ecosystm
With cyber security coming under the spotlight, the security guardrails employed by cloud suppliers could well be a differentiating factor, according to Ecosystm’s Sheedy.
“Security is still a key consideration and Google appears to have the most secure architecture, but it does not mean Azure and AWS are any less secure than what you have in your datacentre,” he says.
In Singapore, major cloud suppliers already comply with major cloud security standards such as the Multi-Tier Cloud Security Singapore Standard (MTCS SS).
First announced in 2014, the MTCS SS drives cloud adoption across industries by providing clarity around the security provisions of cloud service providers. These include cloud governance, infrastructure security, operations management and cloud-specific information security, across three levels of security.
As of 28 May 2018, a total of 130 cloud services from suppliers such as Amazon, Microsoft, Google, Alibaba and SoftLayer have been MTCS SS-certified by accredited certification bodies.
Assessing IaaS performance
There are a variety of metrics that organisations can use to assess the overall performance of public IaaS offerings, including availability, reliability and security.
Derek Wang, chief solution architect at Alibaba Cloud, points out that for compute and memory, reliability and elasticity are the main considerations, while input/output operations per second (IOPS), durability and adaptability need to be prioritised for storage. Good networking performance will also depend on bandwidth, security and load-balancing capabilities.
However, Wang says a better way to assess IaaS performance is to consider how compute, storage and networking work together to solve business problems. “This provides actionable and valuable insights for a business across the entire spectrum, rather than an isolated performance measure in individual areas,” he adds.
In fact, there are benchmarks and tools to help businesses assess the performance of their IaaS offerings that are specific to certain use cases . For example, public IaaS geared towards disaster recovery would have different metrics attached to it versus one geared towards development and testing, according to David Bate, vice-president for cloud at VMware Asia-Pacific and Japan.
The ability to bring new services to market and overall organisational efficiency are also important considerations, Bate says. In this case, project cycle time and costs relative to headcount are metrics to track for some insight.
“As organisations scale their cloud environments and expand the use cases, they may also struggle with how to maximise a multi-cloud model to drive business transformation. They can consider a cloud operations platform like CloudHealth that cuts across AWS, Microsoft Azure and Google Cloud to help analyse and manage cloud cost, usage, security, and performance centrally for native public clouds,” says Bate.
IaaS use cases
Five years ago, it was common to see early adopters of cloud dip their toes by starting with development and testing, followed by backup, disaster recovery and production workloads. But today, major cloud suppliers are seeing large mission critical applications such as SAP being run off their services.
Microsoft’s Bihammar notes that a sizeable proportion of enterprise cloud users today are using IaaS – such as virtual machine-based workloads – as well as applications that are migrated from on-premise to cloud due to hardware refresh cycles in what is commonly known as lift-and-shift migration.
VMware’s Bate sees two distinct categories of IaaS use cases. The first is where an organisation makes a strategic decision to retain its datacentre, but operate under a hybrid cloud model. The other category is where the organisation decides to move enterprise workloads permanently to the cloud and cease running its own datacentre.
“In the first category, organisations are typically motivated by a variety of needs that include scaling out existing capacity, meeting seasonal demands, developing and testing new applications without the need for adding new physical infrastructure, or for disaster recovery,” says Bate.
“In each of these cases, these organisations aim to avoid the significant costs incurred from having their datacentre capacity sitting idle for periods of time.”
Bate says in disaster recovery, for example, high capital and operational costs and complex management can be avoided with IaaS. This use case, also known as disaster recovery as a service (DRaaS), allows enterprises to protect critical data and applications while taking advantage of cloud flexibility and economics to move workloads between on-premise datacentres and public clouds.
The pathway to PaaS
Ecosystm’s Sheedy notes that IaaS, when deployed right, will eventually become business-as-usual capabilities.
“Nearly every company that makes the switch to IaaS will not talk to you about the benefits of IaaS,” Sheedy says. “They will tell you about the benefits of cloud platform capabilities. This is why a decision to embrace IaaS should not be made in isolation – organisations also need to examine the PaaS [platform-as-a-service] capabilities of the cloud provider.
“PaaS is what will drive the ability to innovate and do things differently in the public cloud. PaaS is what will give them the speed and agility that digital businesses demand. IaaS very quickly becomes a hygiene factor – or the building blocks of PaaS-centric capabilities.
“While IaaS was often the reason companies moved to the public cloud, along with a desire to stop spending on their own data centres and computer equipment, what keeps them there and gives them both business and customer benefits is PaaS.”