When Baystate Health embarked on an $8.5 million project to build a new data center three years ago, leaders at...
the hospital system ultimately decided to cut a different path that's still a bit cutting edge in the healthcare realm: They scrapped the project and built a virtualized hyper-converged infrastructure (HCI) environment instead.
As other hospitals look to infrastructure needs, exploring hyper-converged vs. converged approaches reveals varying options to consider.
Baystate, based in Springfield, Mass., needed to provide physicians and staff with a flexible, high-performing technology platform, contain soaring operating and capital costs, and accommodate present and future load demands.
After considering the huge amounts of capital, space, cooling and power that a new data center would require, Baystate opted instead to replace all of its traditional, separate storage, compute and networking elements with a single, software-driven HCI architecture. The setup bundles these elements together with virtualization into a number of appliances, or nodes, said Mike Feld, president and CEO of IT advisory firm VertitechIT, who led the project as Baystate's interim CTO.
Baystate's former CIO had sought a solution that would have predictable, linear costs, so he liked HCI's pay-as-you-grow capability; scaling the architecture simply requires the organization to add more nodes. "He realized that to scale and evolve our IT functionality, we had to make some very serious bets on technology," Feld recalled.
Storage, construction costs drop
The bets paid off. Baystate has cut its storage costs in half, saved $3.5 million in construction costs and decreased its nonstorage hardware costs by 20%, despite migrating such massive platforms to the hyper-converged system as HR, imaging, email, accounting and many of its clinical applications and systems, Feld said. Baystate now has nearly 700 TB of data that it consolidates across three data centers onto more than 40 hyper-converged appliances.
TechTarget's VP of market intelligence goes over converged and hyper-converged EMEA trends based on survey results.
The new IT infrastructure is also deploying nearly 8,000 virtual desktops across the healthcare system, which spans an 80 mile radius. This makes it easier for Baystate's clinicians and administrative personnel to access the network from any device at any time, and to quickly pick up where they left off.
Additionally, the hyper-converged setup requires fewer IT specialists to maintain, so they're finally freed up to tackle such perpetually delayed projects as wayfinding technology and telemedicine.
Explosive growth expected for HCI
Although still considered to be in its introductory stage, the global hyper-converged infrastructure market is exploding. It grew to $1.5 billion in 2016, up from $805 million in 2015. By 2019, it's expected to reach nearly $4 billion, according to research firm IDC.
Healthcare organizations are relatively late adopters of the technology, although their implementations of this technology are increasing as the industry seeks scalable storage, disaster recovery and virtual data access to digitized patient information, according to research by Stratistics MRC.
Hyper-converged vs. converged architectures
Healthcare organizations looking to modernize their IT architectures may be drawn to consider hyper-converged architecture, but it's not the only disruptive infrastructure. Converged systems also hold promise for organizations seeking a superior alternative to the heterogeneous structure of most data centers.
Converged infrastructures (CI) are hardware-based systems that aim to minimize compatibility issues between storage systems, network devices and servers and reduce costs for cooling, power, cabling and floor space. They combine servers, storage and networking from different vendors into one integrated bundle that's certified by the vendor to work together.
Each of these block components can also be separated out and used for their intended purposes, said Eric Slack, a senior analyst with the Evaluator Group. That is, the storage can be separated out and used as functional storage, the server can be separated and used as a server, and so on.
Converged stacks are typically sold either as preconfigured packages or as "cookbook" reference architectures that customers assemble, Slack said. Organizations scale the system by adding individual blocks to the racks.
There are pros and cons to consider when investigating hyper-converged vs. converged options. Here are a few to keep in mind:
Advantages of CI:
- A single vendor handles the integration and provides end-to-end support.
- Organizations experience accelerated and simplified data center deployment with few errors.
- Increased performance occurs that requires fewer resources.
Disadvantages of CI:
- Additional hardware and management software costs may be needed.
- Updates aren't timely -- it can take months before a new processor integrates into a new device due to rigorous testing requirements.
- CI requires a larger physical footprint.
Advantages of HCI:
- It supports incremental scaling.
- There are lower costs due to reducing the number of physical systems that must be purchased.
- The technology is useful for smaller organizations that lack the IT expertise to fine-tune individual components.
Disadvantages of HCI:
- The data center can't select the vendor management software.
- Ripping out old hardware that still has some life in it to accommodate HCI can be unnecessarily expensive.
- Smaller organizations might be unable to find an HCI platform with a small enough configuration.
Learn about data center cooling issues for hyper-convergence
Converged vs. hyper-converged systems and managing virtualization deployments useful in different ways
What flash brings to hospital storage systems