Heathcare providers are under pressure to further reduce costs, offer new capabilities, and explore Cloud options for the future. These pressures require that we revisit the design assumptions of the on premise datacenter to lower costs, reduce complexity, and add capabilities. The platform required to deliver new capabilities and lower the total cost of on premise infrastructure exists today in the Software Defined Datacenter, and customers are adopting it to meet current demands and prepare for the future.
Present Healthcare Pressures
- New capabilities are required e.g. mobility and enhanced security.
- Meaningful Use Stage 3 will put further pressure on providers to control costs in all areas of the business.
- IT spend represents 3-8% of Revenue for a Hospital or Healthcare System.
- CIOs at CHIME have for two years told us that they want to exit the datacenter business in three to seven years.
- Fully migrating to Cloud will take two to five years for the most committed organizations.
- The current standard datacenter architecture is complex and expensive.
Optimizing On Premise Infrastructure
Controlling IT spend is a priority of every Healthcare system and Hospital. Infrastructure and Staff represent significant portions of that budget, and there are significant opportunities to improve the scale of staff while reducing the complexity and cost of the infrastructure via SDDC. Organizations that are embracing the new technologies are realizing savings in all areas of their infrastructure: Storage, Compute, and Networking, and in so doing they are ensuring their competitive strength relative to Public Cloud alternatives.
Complexity is an oft overlooked reality of the current datacenter design. Simplicity is the key to steady operations: reducing the number of moving parts inherently increases the reliability of a system. Application operation requires a delicate confluence of Compute, FibreChannel, Networking, and Storage. These components come from multiple vendors, scale independently, are managed and monitored separately, yet all must work together. This is very difficult to architect, manage, and troubleshoot effectively and overworks a lot of experienced personnel.
Storage presently represents roughly 50% of the annual Infrastructure spend in Healthcare. New software storage solutions deliver the same performance and reduce Storage spend by 30-60%. In a recent Healthcare customer project, they were able to realize a 50% savings on storage while gaining additional compute nodes using software defined storage.
The current virtualization standard of distributed compute nodes backed by a highly resilient and available storage array was a necessary stage in the evolution of the datacenter because of the nature of the workloads: they are special and many cannot be made effectively resilient at the application level, so we rely on the infrastructure layer to deliver the availability. The storage array was the way to do this, and it required the expansion of yet another infrastructure element, the FibreChannel SAN.
By leveraging virtual storage in the compute nodes, significant capital and operating savings are being realized, and due to persistent cost pressures and sound business decision making, it is an emerging standard for efficient on premise architecture.
Compute represents another significant chunk of infrastructure spend, roughly 15-25%. Blades have emerged as the popular option, but it is essential to revisit the reason. It isn’t rack space, host identity management, or any other vendor-specific capability. Blades deliver savings on FibreChannel ports to connect the systems to the storage. There are not significant efficiencies gained from Blades in any other aspect, except as affect the ease of connecting them with FC storage. But what happens if the new storage models do not require FC? The fundamental value proposition of blade architecture erodes and vanishes in favor of lower cost, equivalent capability from rack systems with local disk and software defined storage.
The premier Blade compute vendors are commanding a great deal of spend, but they are not delivering value commensurate with that spend, especially in the face of new distributed storage capabilities that they are not optimal to deliver. Rack mount systems offer greater capability and flexibility for less, and all they need to operate and deliver the same outcomes as current SAN attached designs is power and networking.
Appliance Compute merits a thorough discussion as well. Reducing complexity and cost while adding capability is a challenge, and both ends can be achieved easily via the EVO platform. The EVO platform is a reference hardware architecture with fewer interchangeable parts. We have seen an increase in host instability due to hardware in the last two years. Ever expanding combinations of firmware, storage controllers, network adapters, and drivers have created a hardware ecosystem so large that it is difficult for hardware vendors to test all permutations and combinations of the components. The solution is simple: reference architecture with fewer variables and greater consistency.
EVO rack systems offer everything in one box: Compute, Storage, and Networking. Like the Rack systems, all they need is power and networking, and they deliver all of the capabilities needed by a modern infrastructure platform with more of the capabilities configured and managed in software than ever before.
Current networking architectures are complex and expensive as well, representing 20-30% of infrastructure spend. That cost is in the gear itself and the enhanced security capabilities tied to it. Virtual Networking allows those security policies to be moved up out of the gear, which has significant implications: security policies can be attached to applications and users instead of IPs and ports, and the capabilities of the gear are reduced to efficient packet switching.
By moving the security policies up in the stack, we gain security capabilities there were prohibitively expensive to implement and impractical to maintain, and we allow choice in the gear from many vendors who cost 30-40% less than the dominant communication vendors.
Virtual Networking allows an ecosystem of devices to share in a global policy definition and implementation. We can easily draw boundaries around applications, policies that travel with the workloads as they move about the datacenter and later into the Cloud. Rules are implemented close to the objects and close to the edge. Workloads that cannot talk to the internet can have their packets dropped at the hypervisor; workloads that are in different security zones on the same host can communicate directly without traversing the edge network; and application access can be granted to specific users at the network level – their packets won’t even flow if they are not allowed.
End User Computing and Mobility
The popular way to deliver applications in Healthcare has the same complexity issues as the rest of the datacenter for the same reasons: it leverages expensive compute and storage. Capital costs to deliver SAN attached End User computing infrastructure is frequently upwards of $500 per user. A modern Always on Point of Care infrastructure can deliver a superior Clinical Experience for capital costs less than $250 per user. The operating efficiencies and flexibility offer tremendous value beyond that, but the capital costs are significant and impossible to ignore.
Path to the Cloud: Act Now to Realize the Savings and Prepare for the Future
CIOs at CHIME repeat that exiting the datacenter business is an objective; it is only a question of when, but that transition will take time. With that in mind, there are two realizable short term objectives: invest in the solutions that lower on premise capital and operating cost, and build the operational excellence required to effect a seamless transition to the Cloud with the time comes.
SDDC is the means to deliver on those objectives. Leveraging software defined storage and virtual networking allows compelling savings in storage, compute, and networking. Beyond that, the platform is designed to loosely couple your on premise datacenter with Public Cloud providers to seamlessly migrate workloads along with their operating and security policies with minimal interruption—sometimes no interruption at all. Imagine it: a stretched datacenter with policies defined in software and implemented in the walls of your datacenter and in your portion of a Cloud provider. Administrative control remains with infrastructure and application owners and allows the easiest choice of runtime with the easiest transition.
This is where we are all headed, and the technologies are in use now, today. We can get you there, too.