Subscribe
About

Responding to disruptive change

Today, it’s all about being innovative and outmanoeuvring the competition; companies know that a modern datacentre will help them do exactly that.
Kirsten Doyle
By Kirsten Doyle, ITWeb contributor.
Johannesburg, 10 Nov 2022

Any modern datacentre will employ cloud in some form or another. Its role has been integral in being innovative, remaining competitive and even sustainable in business. This became painfully clear when the pandemic struck. Almost overnight, there was a mass exodus of businesses scurrying to the cloud to enable distributed workforces. Many organisations got up and running rapidly in the public cloud, and its well understood that those that were advanced on their cloud journeys and cloud-ready did infinitely better than those that weren’t.

But is it that simple? Optimising workloads for the cloud takes a lot of time – a luxury Covid-19 didn’t allow – and not every application moved to the cloud was necessarily suited to that environment.

There is more to getting applications running in the cloud than just copying and pasting all the elements, says Paul Spagnoletti, business unit executive: Cloud at iOCO.

“Workloads must be adapted to run in the new environment, which requires changing drivers, updating DNS and IP configurations, and more. Several legacy applications were written with traditional dedicated and closely connected infrastructure in mind, and these application designs resulted in extremely ‘chatty’ applications, which require a lot of data exchange, either to disk or other directly connected devices. When such an application is brought to the cloud, performance can severely lag because cloud infrastructure, by design, works on loosely connected systems.”

What should have been done differently was to ensure that digital transformation projects had a higher level of visibility and urgency in organisations.

Jeff Jack, Dimension Data MEA

In many instances, entire datacentres were moved to the cloud for customers during this time, says Logicalis’CTO, Morne Laubscher. “Applications not suited for cloud without modernisation or fundamental organisational change include operational technology environments and systems that rely on on-premise instrumentation, such as timekeeping systems. Other applications needed major architectural reworking; SAP on Azure is one. We moved traditional workloads to rearchitected managed instances within Azure and we migrated customers from resource-heavy database servers to managed database instances, which are quick wins from a modernisation perspective.”

Spagnoletti adds that IT departments should have access to reliable configuration management database (CMDB) tools that will easily provide useful information about all the hardware and software components of the company and the relationship between those components. “With such tools and services, you’ll not only have a list of all the applications but also figure out how to prioritise them from most to least urgent. Once you have a complete list of the applications, you can review their technical aspects using relevant criteria in order to determine whether they’re suitable for the migration process. If the application is mission-critical, it should have the lowest priority for cloud migration because of the uncertainty of the new infrastructure. For enterprises, it often makes sense to start with the lowest-risk, lowest-value apps – those with minimal customer data and other sensitive information or applications that can benefit from the cloud’s elasticity. Applications that don’t contain sensitive data are good candidates for the cloud; those that do may be a better fit for a hybrid cloud. Systems of record are repositories for some of an organisation’s most sensitive information, including customer data and financial records. Regulatory compliance and governance best practices may, therefore, prevent the migration of this data to the cloud.”

Complexity


For Jeff Jack, principal head of Cloud at Dimension Data MEA, there are logical groupings to be taken into consideration: whether or not the application is of high or low business value or impact to the business; is the software commercially available off-the-shelf (COTS), or bespoke or custom-developed software; and whether or not the software is suitable to take advantage of the elasticity of resources (and thus cost), together with the advanced features that many cloud providers offer. “The most value derived is from high-value, high-suitability COTS or bespoke software moving to the cloud. Companies that were able to migrate and modernise their applications in these categories would have seen the most value when it came to optimising cost, performance, scalability and agility through the pandemic, depending on whether they were needing to scale their business back, up or shifting direction.”

Not cloud-ready

If your datacentre application runs on custom hardware, then it may not be possible to virtualise it. And if it can’t be virtualised, it can’t be re-hosted in the cloud, says Paul Spagnoletti of iOCO. There are various classes of applications requiring custom/proprietary hardware:

  • Applications running on mainframes can’t be migrated to the cloud without modernising or refactoring for the cloud.
  • Applications like Oracle Exadata, Oracle Exalytics and Oracle Exalogic running on engineered systems that can’t be re-hosted.
  • Applications dependent on a physical machine (such as scanners requiring a specific MAC address, apps requiring specialised hardware drivers that can’t be virtualised, apps requiring a USB token to validate the licence).
  • Applications that require purchase of an appliance (for example, load-balancers, and intrusion detection/prevention) may require further investigation to determine if the vendor provides a virtual appliance alternative.
  • Applications running on proprietary operating systems not supported in the cloud aren’t good candidates for re-hosting, either. IBM AIX, HP-UX, macOS are some examples. Like applications running on custom hardware, these applications may require modernising or refactoring for the cloud.

API-driven automation scales networks, virtual machines, and storage at a moment’s notice. That may lead to higher bills initially, but it also empowers you to optimise for performance or cost depending on the needs of each application, says Spagnoletti. Utility pricing in clouds gives you the benefit of ‘pay as you go’,scale up and pay more when demand is high and then scale back afterwards. A common phrase, ‘own the base, rent the peak’ means you have a base level of infrastructure running with the ability to scale up during high-demand periods.

The location and frequency of communication between the application and external systems will also affect the complexity of moving an app, adds Spagnoletti. “Cloud is an excellent option if an application is standalone or only does batch integrations. If you have any data sources or legacy apps to integrate with, migrating to the cloud will not be a simple exercise as it could result in latency issues. Applications with fewer dependencies are usually more suitable candidates for cloud migration.

“If you’ve been running applications on your company’s on-site datacentres, you might have installed other applications that enable the datacentre to work seamlessly with your applications. If this is the case, you might encounter problems during migration because the cloud might not provide the same kind of support your applications have come to rely on. Similarly, if a group of applications in your datacentre is programmed to constantly share information with one another, moving them to the cloud without replicating the same ease of communication might not be feasible.”

Over and above the speed at which organisations needed to react to the sudden shift to work-from-home, they experienced support issues due to a lack of cloud skills in South Africa. According to Jack, most support issues related to the internal ability and capacity of companies to migrate, modernise and operate cloud infrastructure at scale. Most companies that found themselves in this situation looked to external assistance to build, migrate and operate cloud infrastructure.

Cloudreach research reveals 46% of companies say the cloud skills gap either slowed them down or posed an existential crisis to the company, says Spagnoletti. “Business leaders are finding that their cloud transformation initiatives are increasingly hampered by a lack of skilled professionals, with 34% of respondents explaining that the shortage has reduced their ability to operate and launch services.” According to Gartner, the shortage of technical skills in IT staff, particularly in cloud computing, has become a huge challenge, to the point where organisations are losing market share and revenue due to a cloud expertise deficiency that prevents them from gaining the benefits of the cloud and emerging technologies. “There was a shortage of experienced cloud talent before Covid-19, and the pandemic triggered a new wave of demand,” Spagnoletti adds. “The expertise problem is compounded further by the conflicting priorities of keeping the operational lights on while also finding room to innovate. With technologies such as AI and containers advancing, as well as new security protocols that continually need to be put into place, those wanting to reap the biggest rewards from cloud technology need to make a concerted investment into the right partnerships. 

The shortage of technical skills in IT staff, particularly in cloud computing, has become a huge challenge, to the point where organisations are losing market share and revenue due to a cloud expertise deficiency.

Paul Spagnoletti, iOCO

In today’s environment, the boundaries between infrastructure and applications and ‘build’ and ‘operate’ are increasingly blurry. To meet these challenges, organisations need specialised engineering and operations capabilities from their partners and an approach to service delivery that’s more agile and suited to handle both traditional applications and cloud-native operating models.”

Wide impact


One lesson learned from the pandemic is that organisations of all types have begun to trust public cloud environments to host critical workloads, says Spagnoletti. “From support for mobile applications, to collaboration and content management, to business continuity, the security, redundancy, and resilience the public cloud offers provide guarantees that are much greater than what most private datacentres can support.”

Lessons learnt

When asked what lessons have been learnt from the pandemic,Dimension Data’s Jeff Jack lists his top three:

  • Installed hardware capacity is finite and supply chains are brittle; decoupling your capacity needs from physical hardware should be a strong consideration.
  • Network infrastructure that was built to only cater for datacentre to branch traffic flows was left wanting when most employees were at home or when clients moved from stores to ecommerce, for example.
Security boundaries designed to create a moat around organisations didn’t scale well when employees were forced to access applications from home and remote locations, especially when security threats evolved to target remote employees more.

Laubscher says there has been broad adoption in collaboration technologies in the cloud. “If customers had a bigger focus on a greater transformation journey, quicker cross-platform innovations and efficiencies would have been achieved faster than in a hybrid scenario. Aspects such as security, data insights and AI are just a few examples of such innovations that the hyperscalers very successfully democratised. In addition, the global silicon challenges have turned hardware conversations into a major crisis. Many projects are delayed by more than nine months, with ageing infrastructures putting customer businesses at risk.”

What the pandemic highlighted in an obvious and apparent way was that those companies with a high degree of digital transformation and cloud adoption were better off, says Jack. “Thus, what should have been done differently was to ensure that digital transformation projects had a higher level of visibility and urgency in organisations. Now that the worst of the pandemic appears to have passed, the pace of digital transformation should not go back to pre-pandemic levels.”

* This feature was first published in the November edition of ITWeb's Brainstorm magazine.

Share