As with many large enterprises, Hong Kong-based Tal Apparel’s legacy IT infrastructure had held back its business expansion in Asia.
Complex issues in IT infrastructure and a shortage of skilled IT talent in local markets have also made it challenging for the company to maintain its IT facilities in high-growth markets such as Indonesia, Thailand and Vietnam.
To improve efficiency and support further business expansion, Tal Apparel – which runs 11 factories in Asia – began identifying ways to transform its IT operations and embrace a simple, yet scalable infrastructure.
Kai Yuen Kiang, vice-president for IT at Tal Apparel, says the company researched offerings by various suppliers, including hyper-convergence players such as Nutanix.
“Given that legacy vendors often lack the flexibility and agility to respond quickly to changes in business needs, after a year-long, in-depth analysis, we selected Nutanix, which has an outstanding track record in web-scale technologies and hyper-convergence,” said Kiang.
Initiatives such as Tal Apparel’s move to a hyper-converged infrastructure are the envy of organisations that are looking to simplify their traditional IT infrastructure. Many have been saddled with monolithic compute, storage and networking systems that require separate teams to manage and maintain.
“The silos created by traditional datacentre infrastructure often present barriers to change and progress, adding complexity to every step, from ordering to deployment to management,” says Anjan Srinivas, senior director of product management at Nutanix.
“IT does not want to babysit infrastructure and spend 80% of its time managing it. They just want it to operate as expected, provide high availability and to proactively alert and heal when things go wrong. This is what we refer to as invisible infrastructure.”
Turnkey datacentre in a box
While hyper-convergence may signal a return to a mainframe-like architecture, combining storage, compute and networking in a single box, it can mean different things to different people.
Nishchal Khorana, director for cloud and datacentre at Frost and Sullivan Asia Pacific, says that although a hyper-converged system may imply different architectures based on context, conceptually it refers to integrated technologies – servers and storage – that are managed as a single unit via software-based controllers.
“Hyper-converged infrastructure and solutions offer multiple advantages to IT decision-makers, thus driving the adoption of these technologies,” he says. “One of the biggest advantages is scalability while avoiding over-provisioning.”
Besides enabling organisations to right-size the amount of IT resources needed at any point in time – translating to cost savings – hyper-converged systems also offer simpler design, enable faster deployment and reduce administrative overheads, says Khorana. “Further, since compute, storage and network are coupled via software, it enables IT teams to manage fewer suppliers, increasing accountability.”
Nutanix’s Srinivas notes that although some people think hyper-convergence is about bringing storage and compute together, the term “under-represents the enormous impact of running a cloud-like platform on-premise”.
“We built our initial software based on web-scale engineering, allowing any enterprise application to work on our platform that provides compute and storage,” he says. “Then we added layers of virtualisation, networking, automation and operation management.”
Read more about hyper-convergence
- The adoption of converged and hyper-converged infrastructures in Southeast Asia is in its early stages, but growth is expected.
- Hyper-converged storage/compute, such as IBM’s modular xSeries, is posing a threat to standalone SAN/NAS in the datacentre.
- When evaluating your options for a hyper-converged architecture, simplicity, resilience and scalability are some factors that should top your shopping list.
- Forrester rates 12 hyper-converged infrastructure providers according to products, company strategy and market presence, with Nutanix, Simplivity and Pivot3 coming out on top-line.
However, Nutanix’s rival, Simplivity – now owned by Hewlett Packard Enterprise (HPE) – says hyper-convergence offerings that bundle the primary stack – servers, storage, virtualisation and management – do not address the broader needs of the datacentre and so do not go far enough in reducing complexity and cost.
Simplivity touts what it calls “convergence 3.0”, delivered through a single IT stack that runs on commodity x86 resources and scales elastically by adding resources such as compute power, memory, flash storage, hard disk drives and network ports at a ratio that an organisation wants.
“All resources are virtualised and available to the business or infrastructure applications, delivering the promise of a single shared resource pool,” says Brendan Sit, technology evangelist at HPE’s software-defined and cloud group in Asia-Pacific and Japan. “It also enables global federated management of all IT sites and datacentres from a single management point, including the ability to efficiently mobilise data across sites.”
Put simply, hyper-convergence brings a cloud-like infrastructure on-premise and into a datacentre, addressing the latency and data sovereignty concerns some enterprises may have with public clouds. This makes it possible to deploy applications on-premise quickly, with the flexibility and scalability often associated with the cloud.
“A hyper-converged system is essentially a turnkey datacentre in a box that incorporates the entire IT stack and data services that can be easily procured and deployed in a matter of minutes as opposed to hours or even days,” says Sit.
Early hyper-converged deployments focused on specific applications, such as virtual desktop infrastructure. But today, more enterprises are running a broader set of tier-1 enterprise workloads such as SAP, Oracle, Microsoft Exchange, SQL Server, SharePoint and even Splunk, a security information and event management system, at least according to Nutanix.
Fuelling the growing deployment of mission-critical applications on hyper-converged infrastructure is the wider adoption – and falling cost – of flash storage.
“The introduction of flash has meant that use cases can be extended to virtual server infrastructure deployments hosting a range of applications,” says Courtney Dodds, Cisco’s managing director for datacentre in Asia Pacific and Japan.
Dodds says Cisco HyperFlex goes one step further by introducing a purpose-built hyper-converged infrastructure filesystem, with an optimised and integrated network fabric for running business-critical workloads such as disaster recovery.
But Michael Alp, vice-president of Pure Storage in Asia Pacific and Japan, says the ease of provisioning and managing hyper-converged systems limits scalability, flexibility and performance, behind the layers of virtualisation, integration and automation. As such, he still sees most hyper-converged systems being used to run single and smaller workloads, where less mission-critical data is stored, and performance and scale are not required.
“When it comes to the larger data-intensive workloads, most customers still want to separate compute from storage and get the best scale, cost-efficiency and functionality from their individual storage, server, service automation and switching components,” he says. “This is why Pure Storage is focused on more business-critical and performance-hungry workloads.”
Hyper-convergence is buzzing
Nonetheless, the adoption rate of hyper-convergence in the APAC region is set to grow rapidly. Gartner recently predicted it would grow from $371.5m in 2014 to nearly $5bn by 2019.
“In APAC, hyper-convergence is creating a huge buzz, similar to what we have seen around all-flash storage arrays,” says HPE’s Sit. “HPE believes that the growth in APAC will be in line with worldwide growth predictions.”
But it is not just the larger enterprises that are embracing hyper-converged infrastructure. Smaller and mid-sized companies are also doing so to replace their datacentre environment.
Sit says: “Hyper-convergence is especially compelling among businesses with remote locations or branch offices, which will benefit from standardised systems that are pre-integrated for efficient deployment, easy to manage and scale. To keep pace with growing and sometimes unpredictable workloads, remote sites need systems that can scale quickly and seamlessly.”
To maintain business continuity, these organisations need systems that allow centralised backup, recovery and replication of data, following standard configurations and corporate processes. Hyper-converged systems meet all these needs, says Sit.
He also points out a strong case for hyper-convergence to support lines of business within large organisations, such as a typical finance department that routinely generates reports at the end of each quarter.
“As the data grows, the reports take longer and longer to process,” says Sit. “The finance department may also have new projects in the pipeline, such as an advanced accounting system that needs to be rolled out within the quarter.”
In such cases, Sit says a hyper-converged system is ideal to support new applications or services, because it can be ordered easily and set up quickly with pre-scripted configuration tools.
“The ability to quickly roll out a fully functioning IT environment could allow the company to deploy its new accounting system before the end of the quarter, generate reports in a timely manner and avoid the end-of-quarter chaos of the past,” he says.