Momchil Michalov, VP, Storage Technologies, Citrix
Modern datacenter architecture allows us to utilize hardware to empower software. This shift, which has been called the software-defined datacenter (SDDC), allows us to create a more fluid, dynamic, and responsive infrastructure. Such an infrastructure provides the flexibility to quickly adjust to meet rapidly changing workloads, data streams, and computing demands.
"We need to focus on and deliver a consistent and responsive workload user experience across many end-user devices, including workstations, laptops, smartphones, and tablets"
Virtual workspace delivery through virtual desktop infrastructure (VDI) and application delivery, where end users access applications running in the cloud or in the datacenter, greatly benefits from this type of flexibility.
By delivering the workspace virtually, we enable organizations to improve productivity, and workload efficiency empowering the workforce to access their applications and data from anywhere, on any device. Yet, to do that, we also need to focus on and deliver a consistent and responsive workload user experience across many end-user devices, including workstations, laptops, smartphones, and tablets.
The Unique Storage Challenges Involved in VDI and Workspace Delivery
The lowest common denominator in most IT deployments is storage. Since VDI first emerged, storage performance has been a key factor in delivering a quality end-user experience. Yet for a long time, storage has been a black voodoo box for IT. It has been very expensive to procure and manage, and it is frequently one of the most complex, yet poorly performing infrastructure elements in the customer’s datacenter.
However, storage holds one of the most precious resources in the organization—user and application data, which are the keys to productivity and output when delivering virtual workspaces.
VDI and workspace deployments have been significantly hindered by the cost, complexity, and performance of storage. For virtual workspace delivery to provide the full value businesses need, the industry must not only address performance challenges, but must also reduce the per-session cost and deliver a better user experience.
Redefining the Storage Architecture to Reduce the Workspace Delivery Cost and Complexity
For a long time, storage has been a monolithic and expensive piece of infrastructure lagging behind more modular server, networking, and application architectures. In addition, server virtualization and application delivery have added fluidity and dynamic expansion and migration to these layers. This cannot be matched by conventional storage infrastructures.
To make matters even more difficult for customers, an average storage array lives in the customer datacenter for 10 years. As a result, new storage architectures must enable these capabilities, while augmenting current infrastructure, rather than requiring a rip and replace approach.
New storage architectures should be modular, fully distributed and should take advantage of industry-standard components, coupling them into elastic storage, data and compute systems through software. These systems can then be used to deliver elastic VDI and workspace delivery seamlessly, with a focus on defining and guaranteeing the user experience while eliminating infrastructure and storage islands.
The notion of separate compute and storage islands for workloads and data needs to be consolidated into an SDDC architecture where CIOs and IT departments can enable services, capabilities, and quality of service (QoS) through software that is deployed on generic hardware. Over time, we will see the workload capability and coverage from these new systems expand dramatically.
Defining the architecture
The new software-defined storage architecture allows us to decouple the data and compute resources from the underlying hardware. It enables us to layer the necessary services on an as-needed basis where key capabilities are a function of unified software across heterogeneous hardware versus the current significant limitation and cost burden of homogenous hardware. It enables much faster innovation, deployment and updates of functionality but also addresses a key challenge for IT teams – when and where to run workloads for best performance and cost. By moving the enablement process into a distributed software stack, customers can utilize their infrastructure as a service to span across their datacenters, eliminating the dilemma of deploying in the cloud or on premise.
Implementing the architecture
One size does not fit all. Customers need the ability to utilize existing on-premise hardware along with resources deployed in private clouds, with infrastructure as a service, and many physical and virtual platforms. While single vendors may continue to work on providing the complete stack, it’s important to take into account the diversity of infrastructures and needs within many enterprises. There is tremendous demand for an “any-ness” like architecture that can integrate multiple types of storage, networking, and compute to be provisioned and orchestrated in a more open way through a host-based OS-independent stack.
Key requirements for the architecture
While the software-defined datacenter and workspace are becoming established in the market, customers need to look ahead to further enable this flexibility and elasticity in their environments. CIOs continue to face tight budgets and organizations that demand more services with ever-shrinking deployment time frames.
Balancing provisioning and economics is tricky, so the ability to have a single control plane across workloads and infrastructure is fundamental. The right mixture of physical, virtual and cloud resources coupled with the right network infrastructure and centralized unified deployment, as well as services and management platforms that can span geographic locations can overcome both budget and time challenges.
Streamlining Workspace Delivery with a Software-defined Storage Architecture
We have arrived at a point where hardware can be used to enable software as opposed to the other way around. The elimination of proprietary hardware helps dramatically reduce deployment capital expenditures, and the migration of management and provisioning in standard OS platforms and tools significantly improves operational expenses and efficiency.
Through software, we can deploy and repurpose the same infrastructure for multiple applications and workloads while significantly reducing time to value and increasing productivity. Instead of deploying a proprietary stack to enable its own availability, we can deploy stacks that enable the infrastructure availability. More importantly, this provides a holistic workload approach where VDI, applications and ultimately businesses data are highly available, scalable, and elastic, which dramatically improves productivity.