All posts by joshepstein

Assessing data centre strategies for cloud-scale software

The rapid advancement of cloud-scale software is driving the digital transformation impacting nearly every facet of our life. The ways we work, communicate, navigate, travel, shop, manage our money, access healthcare, interact with things and places are extremely different from only five years ago.

There are many underlying technologies that enable this transformation – but none as profound as the rise of cloud-based software applications that we interact with throughout our day. 

IDC notes that cloud software accounts for nearly one-third of the $400bn+ software market and is driving virtually all of the organic growth in the industry. This category of technology includes software as a service offerings; eCommerce sites; fintech and health tech apps. It includes emerging segments including IoT-enabled applications and AI-enabled lifestyle applications like autonomous vehicles and smart homes. 

Data centre strategies for cloud software are different than those that support traditional software applications. Whether hosted in public IaaS or in a private or managed private cloud, it is important to recognise it requires several characteristics that make modern cloud software unique from software hosted locally or in an enterprise data centre.

Application multi-tenancy

Cloud software is by nature multi-tenant – supporting many users from different organisations on shared infrastructure but with logical divisions between different users’ private data.  Supporting consistent user experience means application infrastructure must adapt to surges in activity across the full user base.

Agile development and DevOps

One of the benefits of managing software in a central cloud-scale data centre is the ability to release new functionality more frequently. The rise of agile development and DevOps lets organisations role out new software releases weekly or even more frequently. Infrastructure bust be adaptable in order to not be a bottleneck in introducing new functionality.

Scale

Modern software-driven businesses are built on the assumption they will scale. More users, more devices, more data. This requires data centre infrastructure strategies that can scale seamlessly and predictably with the businesses they support. It also means sensible cost strategies. One of the goals of any business is to realise economies of scale as they grow. One challenge with public IaaS offerings is that cloud storage and computing bills rise nearly linearly with the business.  Ideally data centre infrastructure will continue to deliver performance and responsiveness to applications as they scale while improving cost-efficiency.

Powerful analytics

Probably the most significant requirements for modern cloud software is the growing importance of sophisticated analytics-based functionality. Real-time analytics are at the core of personalised user experiences in ecommerce platforms. Analytics are behind the sophisticated business insights at the core of modern enterprise applications. Analytics drive the machine learning important for AI enabled applications. And analytics power the new semantic interfaces and chatbots that are re-shaping how users interact with software. 

Data centres need to be able to manage these analytics on the scale of traditional high performance computing applications. Scale-out, all-flash storage has been a key enabler to delivering analytics performance at scale. Emerging technologies based on low-latency NVMe networking and advanced solid-state memory designs promise to enable even greater levels of infrastructure performance than available today.

As product, DevOps, IT, and finance teams chart out data centre strategies to support cloud scale software applications, they must navigate an increasingly complex ecosystem of technologies, service providers, and architectural paradigms. The next generation of technology and service providers will need to help reconstruct the data centre technology stack to support the needs of modern, cloud software.

A guide: How the rise of SaaS and cloud influences the modern data centre

Cloud delivery of enterprise applications is rapidly transforming the software industry, IT organisations, and the modern data centre. The as-a-service model for delivering advanced software functionality has moved into mainstream acceptance. IDC projects the cloud software market to grow to $151.6 billion by 2020 with a five-year compound annual growth rate (CAGR) of 18.6% – far exceeding the growth of traditional software.

Enterprise IT organisations are embracing cloud software for good reason. SaaS versions of enterprise applications introduce new innovations faster than traditional software. Faster public networks, pervasive mobile devices, and modern development paradigms enable excellent user experience. Powerful cloud-based data centres serve up sophisticated real-time analytics functionality to users accessing applications from a variety of devices.  And probably most significantly, the adoption of SaaS frees up IT resources and simplifies enterprise data centres. 

The adoption of SaaS and cloud based software is a cross industry phenomenon. Health tech, fintech, retail tech offerings are emerging alongside enterprise applications-as-a-service (e.g. ERP, CRM, HR, accounting).  SaaS versions of enterprise applications to smaller businesses that never had the option to host complicated applications on premises.  Looking ahead, the rise of IoT enabled businesses – including smart industry, connected health, and smart city oriented companies – will also be delivered through as-a-service offerings.   These applications all share common infrastructure requirements.  This growing demand is giving rise to a new generation of data centre technologies that address the needs of as-a-service businesses.  

Infrastructure needs for ‘as-a-service’ businesses

Modern SaaS and cloud-based software place new demands on data centre infrastructure.  Some demands are simple evolutions of the requirements for traditional enterprise application infrastructure. Other are unique to the as-a-Service model where a single application instance is supporting many different organizations.

Modern SaaS businesses compete on user experience and functionality. Database platforms, server, and storage infrastructure must deliver the performance needed to deliver a positive user experience.   Many modern applications incorporate sophisticated real-time analytics functionality.  For instance, eCommerce sites may want to make real-time recommendations as part of a personalised shopping experience.  Real-time analytics put complicated workload requirements on underlying infrastructure and require unique approaches to maintaining consistently high performance. 

Part of staying competitive on user experience and functionality is the ability to constantly introduce new versions of a software offering.  One of the benefits of cloud software is the ability to roll out new software versions to all users extremely efficiently.  Modern software development organisations have embraced this concept with Agile development methodologies and by incorporating DevOps thinking into their organisations.  As new functionality is rolled out continually, the underlying data centre infrastructure must be as agile as the application. Rigid management paradigms and complexity have no place in the modern data centre.

Probably the most significant requirement for as-a-service infrastructure is the need for scalability. Modern SaaS business models are based on the principle of scale.  Scalable infrastructure strategies are critical as underlying business growth drives growth in users, in devices, in transactions, and in raw data.   Infrastructure solutions that cost-effectively deliver performance, simplicity, and reliability for a single enterprise often do not support the scalability needs of SaaS and cloud software.

Transformative technologies for the ‘as-a-service’ world

Service virtualization transformed the economics of data centre management, paving the way for the as-a-service world.   Beyond that, several infrastructure technologies are continuing to transform the modern data centre.

Solid state storage: Enterprise storage platforms built for solid state disk technologies are rapidly taking share from traditional hard-drive based arrays.   All-flash arrays that leverage the latest solid-state technologies and incorporate efficient data reduction technologies enable a step-change in performance with costs competitive with traditional storage.

Software defined data centres: Across the data centre stack (including servers, networking, and storage), software defined principles enable a new class of highly flexible, highly cost-efficient infrastructures solutions.  Custom hardware based solutions are unable to innovate at the rate of software-defined solutions that can ride the commodity hardware curves. 

Convergence: High speed data centre networking technologies including NVMeF will transform the notion of shared storage infrastructure. By connecting server resources to shared storage with this extremely low latency interconnect, data centres will achieve a new level of performance and flexibility.

New database paradigms:   As the server and storage layers of the data centre transform, database and application development principles will shift to gain full advantage.  Traditional RDBMS technologies will be augmented or supplanted with alternative database technologies like NoSQL/NewSQL to more efficiently deliver advanced functionality like real-time analytics. 

‘As-a-service’ data centre winners and losers

The rapid growth of SaaS and cloud software is not reducing the market for data centre technologies, it is driving a massive shift in who is buying these solutions and what their buying criteria are.  Data centre technologies that have been successful on premises, will not necessarily have the same success in the as-a-service world.  Legacy solutions saddled with rigid hardware based architectures are unlikely to keep pace with more agile, software based solutions.   Technologies with proprietary interfaces will lose to those that offer developers access through open standards.  Complex technologies requiring specialist support resources will give way to simpler solutions that can be managed by generalists.  

The as-a-service data centre will be architected for scale, for simplicity, and for the applications that will drive the future of digital business and digital lifestyle.

The four key storage drivers for SaaS success

(c)iStock.com/livecal

Software as a service (SaaS) and cloud delivery models are disrupting the world of traditional enterprise software.  As IT organisations embrace agile infrastructure strategies to keep pace with the rapid change of the businesses they support, SaaS delivered applications have a clear advantage over traditional on-premises enterprise apps.  As the forces of digital transformation reshape enterprise IT for the on-demand world, SaaS applications will be a driving force in reshaping how modern businesses function.

Choosing SaaS

SaaS is a more efficient way to deliver powerful functionality and positive experiences to business users, as well as radically simplifies deployment. Users of SaaS solutions also don’t have to install and maintain servers, databases or scale data center infrastructure to add more storage or compute power – the SaaS vendors handle all of this for them. The SaaS model collapses the entire application delivery infrastructure into the software being delivered, which is why SaaS is so appealing to CIOs and line of business executives.

Some SaaS providers turn to public cloud vendors, though others may choose to build their own cloud infrastructure to optimise application delivery infrastructure as part of their core competence. Furthermore, providers want to optimise this to ensure seamless delivery when it comes to issues around customer experience, real-time analytics or privacy safeguards. For SaaS vendors, solid, flexible infrastructure can be a competitive advantage in delivering mixed workloads.

The four keys to success

In order for SaaS vendors’ infrastructure strategies to differ from enterprise IT and ensure it impacts the effectiveness of the service – the application “experience” they offer to end users –vendors should consider these four SaaS business drivers when it comes to impacting application infrastructure decisions:

  • Customer retention is as important as a new customer acquisition – Because SaaS users aren’t invested in infrastructure and perpetual license outlays, the “switching cost” is low. As a result, SaaS vendors must provide consistently high customer experience in every area, including great uptime, superior application performance and general account management.
  • SaaS vendors most grow to succeed – Vendors’ business models are based on low barriers to start (e.g. free trials) and network effects that deliver more value as user base grows. This means that SaaS companies need infrastructure that scales easily as they add customers, users, devices and data growth.
  • Deliver real time analytics and decision support – SaaS solutions today are doing far more than just transactions—they need to manage real-time analytics. The era when online analytic processing (OLAP) was a batch function performed by specialists with niche data warehouse tools is now giving way to SaaS solutions that embed personalisation, intelligence and other analytics as part of the user experience. This requires infrastructure that can deliver consistently high performance and user experience under dynamic workload environments.
  • Ability to scale – The cost of application delivery infrastructure is effectively the cost of goods sold (COGS) for a SaaS company. As a result, the ability to scale infrastructure in a cost efficient way is vital to profitability.

These four key business drivers map back to capabilities needed from cloud infrastructure. Scalability is table stakes – any SaaS vendor that chooses to build out its own infrastructure must ensure it can quickly scale its application and performance needs, even as its application workload profile evolves to include more rapid analytics and personalisation.

The infrastructure stack for any vendor building its own cloud has multiple layers – from virtualisation to security and disaster recovery capabilities – but at the end of the day, its foundation is storage. As a result, SaaS providers must consider these infrastructure requirements when it comes to choosing a storage technology approach.

The shift to all flash storage?

To power the high performance needs of the modern SaaS business, all flash storage is the natural technology selection. However, beyond choosing all flash, SaaS providers also need agile storage solutions that can seamlessly manage growth in data and users with a cost model that supports their business. Furthermore, storage solutions must evolve with the constantly shifting requirements of the applications they support.

The ability of a storage platform to handle SaaS solutions with real time analytics is the biggest concern. The platform will need an architecture that performs at the highest level for both transactional and analytics workloads. Not every storage array can deliver consistently high performance under both types of workloads. In order to deliver consistently high user experience in a cost effective manner, it’s essential that the entire infrastructure stack – from the bottom storage layer up – is tuned for real-time analysis, as well as transaction processing.

Conclusion

While much has been made of the SaaS business model – subscription-based solutions that offload the infrastructure concerns for users – the future for what will set one SaaS solution ahead of its competitors is likely to be how well it can deliver differentiated functionality like advanced analytics. Most SaaS providers have adopted agile development practices and incorporated DevOps into their thinking to stay responsive to customer needs and maintain a competitive advantage. At the same time, deploying an agile infrastructure strategy that incorporates flexible storage solutions is essential, so that it can evolve with changing demands of the application and market.

SaaS is already eating up the traditional software delivery model, and in order to succeed, vendors must focus on embedding real-time analytics to deliver a strong user experience. At the end of day, this means SaaS infrastructure not only has to scale, it must also handle complex workloads in a way that will support the most competitive end-user experience. The winners in the age of SaaS will make strategic infrastructure choices that best support their fundamental offering and business model.