Five critical challenges to building cloud services

By Gary Lyng Cloud architects looking to transform their organisations should look at how top cloud providers support a range of workloads, use cases, and I/O patterns. No matter where we look in industry, the sector with the highest appetite for storage and the fastest growth rate is the cloud services marketplace. More and more enterprise workloads are moving towards cloud providers. With the growing versatility of Platform as a Service (PaaS), Infrastructure as a Service (IaaS) and Software as a Service (SaaS) offerings, come new end user and IT expectations for business agility, self-service, performance and infinite scalability — and those bring on great challenges for cloud architects to keep up with demand. As per the latest Gartner report, the IaaS segment will remain the fastest-growing segment in 2016, forecast to reach $448.9 million. Building a cloud infrastructure to accommodate enterprise and mobile workloads can be daunting. It is an entirely different beast than the traditional, on-premises siloed storage approach enterprises have traditionally implemented. And there’s no one recipe. Each enterprise customer will have different needs and goals depending on their expectations from the cloud. Yet there are critical architectural components you need to keep in mind to make your cloud project successful: Range of workloads: When designing robust cloud services, underlying storage to support such services needs to be versatile for the varying random I/O and data throughput requirements. The type of data and how we use it is changing. Analytics is giving way to Predictive Analytics. In our globalised economy, it is no longer acceptable to make decisions from historical data. Business executives are now pushing for better and up-to-date predictive analytics. An entire rework of cloud infrastructure will be needed to carry out cloud-based analytics. DevOps culture and continuous development: If you’re a cloud service provider, your customers expect the infrastructure to provide a PaaS environment for further development. Providing a PaaS for DevOps will entail test/dev environments and highly-reliable production infrastructure. Customer Experiences: Customer satisfaction is only as good as the lowest network and storage latency within your infrastructure. Architecting cloud services with predictable and fastest response times means flash storage that has been tuned for specific latency-sensitive workloads. This is key to QoS (Quality of Service) or better stated ‘Consistent Quality of Service’ Scale: With the exponential growth of data, cloud architectures require a new approach that will enable them to scale storage without the undue costs of compute. Storage need not be tied to computing as it scales. Disaggregated storage is an approach to balance the demand for storage bandwidth without the unnecessary costs of CPU/RAM. How IaaS providers power their cloud So how do top cloud providers build infrastructure to support a range of workloads, use cases, and I/O patterns? The reality is that more workloads are moving to flash every day. Here are three reasons why: Performance Unlike traditional spinning media, flash can be designed in ways so it can be heavily workload dependent, with performance variance over time based on block size, random versus sequential I/O, read versus write patterns, temporality, and data throughput requirements. Flash turned the industry around: no longer is the workload designed to fit the storage device limitations, instead flash can be designed to fit the given workload. Persistent storage tiers Persistent tiers both flash and HDD, in cloud infrastructure must satisfy consistent throughput, latency, and performance isolation. Performance must be predictable at any given time, and persistent across the entire given tier. Unlike any other storage medium before, enterprise flash guarantees predictable performance and even a fully predictable lifetime tying back to providing QoS. Innovative use cases Cloud services need to deliver an infrastructure ready for existing on-premises workloads, but the real value appears when we start to talk about innovative new use cases that would have been impossible without the cloud. Database-as-a-Service, analytics in the cloud (e.g., Hadoop-as-a-Service), DevOps environments, Media Rendering as a service, and newly developed applications supporting Internet of Things (IoT) – the possibilities are infinite. The author is Senior Director Marketing & Strategy, Data Center Solutions, SanDisk