Is cloud computing delivering on what it has promised?
Many organisations see the cloud as the ultimate solution – the promised land that delivers everything. And who can blame them? Cloud service providers are pushing the model hard. In fact, the worldwide public cloud services market is projected to grow by 17.3 per cent in 2019 to $US206.2 billion, according to Gartner.
But despite its heavily hyped benefits, cloud computing may not be the answer for some businesses. While one company may see cost savings, scalability and availability benefits from moving to the cloud, the computing model, for various reasons, may not suit another.
Senior tech executives gathered in Melbourne recently to discuss the technical and business challenges driving their move to the cloud, at a lunch sponsored by cloud service provider, Oreta.
Its managing director, Sachin Verma, provided some straightforward advice to organisations in the process of deciding which services to move to the cloud and those that should be kept on-premise under a hybrid model.
“First and foremost, focus on your business. How do you operate? Where are your employees or customers and how do people interact with your applications?” he says.
“Then look at your operating tempo. Are you seasonal with larger peaks and troughs or fairly static throughout the year? Then look at your application stack and understand how it operates and what you want in the future. Will you be expanding into new markets, engaging in mergers and acquisitions or expecting mostly organic growth? All these components need to be understood before any decisions can be confidently made about the relative suitability of any future location.”
Some organisations, according to Verma, aren’t considering these questions and don’t understanding the relative complexity of public cloud spend.
“Cloud can offer elasticity, higher availability and flexibility, but the cost for this capability is sometimes overlooked or misunderstood.
“A prime example is storage. If your budget is based on the cheapest, lowest performance storage price but you need high input output per second (IOPS) to meet your application performance, your price will be significantly higher for the same footprint,” he says.
McMillan Shakespeare (MMSG) group executive, digital and business technology, Dan Giesen, says the financial services organisation conducts a detailed discovery of the services it wants to move to the cloud.
“We then make a call based on whether we believe that using the public cloud will deliver the level of services we are looking for and at a similar cost to our current [infrastructure].
“Cloud providers are delivering technically on what they say they will but it is evident that the costs associated with public cloud computing are not meeting expectations and, as a result, people are ‘pulling back’ from their full cloud. This is exacerbated by limited technical expertise in cloud technologies in the market at this stage,” he says.
APT Travel Group chief information officer, Guy Ballard, says the primary considerations for a move to the public cloud include the application architecture, functional and interface requirements (and the skills needed for them), and the geographic spread of the user base. Other considerations include concerns over data sovereignty and privacy, and the total cost of ownership of the environment.
“In terms of application architecture and functional requirements, our preference is not just to ‘lift and shift’ an application into a public cloud container,” he says.
“Rather, we want the code base to leverage the cloud platform for elements that we can’t cost-effectively develop ourselves on-premise or through private cloud/hosted VM environments. We do have a significant hybrid challenge in refactoring the interfaces between applications we would move to public cloud and their legacy partners remaining in the on-premise of hosting locations.”
Ballard says while many SaaS offerings are ‘public cloud interface ready’, there are many proprietary complexities involved in making them connect efficiently to hybrid applications.
“Our internal resourcing is heavily focused on legacy applications that are hosted on-premise or in an IaaS environment so our cloud skills base is still in its infancy. Formally expanding our skills is a precursor we will need before we make any significant cloud moves.
“The global geographic spread of our organisation means it is difficult to deliver adequate service level and a consistent user experience from an on-premise or local hosting environment. When an application has been used globally but controlled from our central service desk, it immediately makes itself attractive as a cloud candidate,” he says.
Headspace executive director, digital and technology services, Anna Hall, says the youth mental health organisation generally creates a business case to determine if there’s value in moving to cloud services.
“Cloud often makes sense from our perspective as our funding agreements usually span only two years so any purchased infrastructure has to be paid off or written off in that period,” Hall says.
“Cloud security is also often pretty good in comparison to what not-for-profits can afford from small datacentre providers.”
A senior technology executive from a Victorian government department says that a move to the public cloud is generally a decision to better service the customer.
“This means having the website one point closer to the customer without all the infrastructure it would have to cross to get to our application. These added ‘pieces’ are also added points of failure,” he says.
“The next reason [to move to public cloud] might be putting our web presence into a service that takes care of the infrastructure, hosting, network and extensibility. Cost is generally not the consideration.
“As for [cloud] providers delivering, we don’t expect much of them. It’s generally the apps (code) or data that can hiccup and that really is our responsibility unless it’s a Software-as-a-Service [SaaS] application.”
Cutting through the cloud hype
Attendees discussed how they cut through the hype around the benefits of cloud computing, particularly that related to cost savings and scalability improvements.
MMSG’s Giesen-White says organisations should listen to the “war stories” of their peers.
“Talk to organisations that have gone into the cloud and then backed out. If you can find a good partner with a strong understanding of hybrid environments and the benefits of each of the approaches, ask them what their experiences have been,” he says.
“I know that high availability is something that a large number of organisations have always wanted but rarely been able to afford [maybe without completely understanding what it means]. I am sure that cloud solves the problem completely but it is an option. Elasticity is not a significant requirement for us.”
APT Touring’s Ballard says the company has traditionally been a CAPEX-focused organisation where the cost of storage is relatively cheap.
“We have already experienced significant growth in storage costs for some SaaS applications. The potential cost of storage is a key area of caution for us when considering public cloud and helps us cut through the hype,” he says.
“For some global applications, we need to raise the level of performance and user experience. While challenges around interfaces, storage costs and data sovereignty would need to be answered, the cloud providers will make that more achievable than having to establish a mesh of localised instances.”
The tech executive from a Victorian government department claims cloud isn’t cheaper because “unless you take every last bit of kit from onsite and put it in the cloud, you now have to manage two environments.
“If it cost you a dollar to manage onsite and a dollar to manage offsite, it generally adds up to $2.50. When people moved from mainframe to client/server [infrastructure], it didn’t get cheaper. You had to hire sysadmins and manage the network in between and you spawned off copies of data to ‘shadow IT’ groups. That’s a hidden cost.”
Greenstone Financial Services chief information officer, Sanjeev Gupta, says achieving cost savings from cloud services really depends on the workload.
“If you have a consistent workload, the cost savings do not really stack up,” he says. “The scalability, the flexibility and the agility that a public cloud can deliver is a bigger benefit than a cost saving. Not all workloads need the elasticity [that cloud provides],” he says.
Headspace’s Hall says her organisation does not really use the elasticity offered by cloud services.
“Once we put services in, they pretty much stay static until we no longer provide the service, which would be tied to a funding agreement,” she says.
Meanwhile, The Heart Foundation chief information officer, Shane Riddle says cutting through the hype is difficult as “everybody has the best solution”.
“You need to use common sense as marketing hype is generally ahead of the practical application – you need to understand your workloads and bottlenecks.
“One of the great things with cloud computing is that you can test without committing everything. This is not possible in all scenarios but generally you can create a pilot to give you an idea. I use my peer network to assist in sounding out ideas and learn from others that may have gone through a similar experience to pick up the pitfalls to watch out for and to test,” he says.
The legacy challenge
When moving to the cloud, successfully integrating SaaS applications with existing on-premise cloud software is vital and may make the difference between success or failure of cloud projects. And configuring these multiple applications to share data in the cloud is key.
So what do organisations need to do to ensure systems talk to each other?
Fortunately, Greenstone Financial Services’ legacy applications are less than 10 years old and are mostly developed in-house, says Gupta.
“We have access to the source code and we have complete control over these applications. We are able to make the necessary changes to integrate these applications as needed,” he says.
The old saying of, ‘how do you eat an elephant? One bit at a time’ is a great way to look at this, according to Oreta’s Verma.
“API gateways such as Apigee allow for multiple connectivity options into legacy environments. Once the initial connectivity piece is established, it is a matter of chunking down the transformation into stages. Due to the integration capability, you no longer need to make large scale commitments to unrealistic timelines and budgets,” he says.