As budgets continue to tighten, IT leaders are trying to figure out how to extend the life of IT infrastructure so they can reduce spending. At the same time, you must keep up with demands for improving performance and meeting capacity expectations. You need to get more from the current hardware.
There is a process to extending the life of IT Infrastructure that can enable you to reduce costs and maintain application and data availability. It can do this while simplifying IT, which becomes more critical since you may also face a hiring freeze.
Get More From Your Existing Hardware
When an IT vendor approaches you to “help” you cut costs, they first want you to buy more hardware. They want to replace your current hardware, even though it is still working and likely meet most of your performance and capacity demands. Then through the magic of a spreadsheet, they will show you a return on investment (ROI) where that investment pays for itself in 27 seconds. If you spend five or six figures on new hardware, there is no way you will see a fast return on your investment. Getting a fast ROI requires leveraging the existing hardware with a more optimized software platform and then only adding a few hardware components to fill in performance or capacity gaps.
The key to getting more from your existing hardware is to look at the software it is running. Most data centers have three groups of infrastructure that they need to extend their life of; computing, storage, and networking. IT typically buys these as standalone components using a three-tier architecture or bundles using a hyperconverged infrastructure (HCI) solution.
Extend the Life of Computing
In most data centers, computing is managed by a hypervisor for virtual environments or an operating system for bare metal environments. Organizations have bare metal environments, not by choice but because the overhead of the hypervisor precludes the applications running them as virtual machines (VM). The overhead from virtualizing an application will vary from application to application and hypervisor to hypervisor. The overhead of the hypervisor forces organizations to buy far more computing power than they need to compensate for this overhead.
As we discuss in our article “The HCI Resource Efficiency Problem,” this overhead is called the virtualization tax. HCI exacerbates the problem because its hypervisor needs to run storage and networking services as hypervisor-dependent virtual machines (VM).
Extend Computing in Small Increments
What if the server investment is more than four or five years old? In that case, the organization may be at a point where the computing capabilities of its infrastructure can’t keep pace with the last four to five years of growth. In most cases, only a few applications need new servers with more powerful processors. It doesn’t make sense then to “sweep the floor.” You should be able to add the few servers you need in small increments and integrate them into the operating platform. If the new solution can reduce hypervisor overhead, the remaining applications will run on the existing servers for years.
Neither solution, the three-tier architecture or HCI, can fix the heart of the problem, the hypervisor. Three-tier architectures have no control over it, and HCI typically borrows it. HCI also can’t support a mixture of existing server hardware and a few new servers with the latest generation processor.
Extending Server Life with UCI
Extending the life of the existing computing investment requires a new data center operating platform. It needs to go beyond HCI and deliver a highly optimized ultraconverged infrastructure (UCI) like VergeOS, which rotates the traditional stack (computing, networking, and storage) into a single linear plane, integrating that stack into a single code base for maximum efficiency and simplicity.
VergeOS provides a more efficient hypervisor that can breathe new life into existing virtualized applications and enables migrating those formally bare metal systems into the same UCI. It also supports easy migration from legacy hypervisors like VMware, Hyper-V, and KVM while supporting all base-level operating systems and environments like Windows, Linux, and Docker/Kubernetes.
Finally, VergeOS supports a mixture of different computing node types. First, using the existing hardware, most applications will experience better performance using VergeOS. Then as part of an incremental upgrade, the organization can add a few nodes with Intel, AMD processors, or Nvidia GPUs, no matter the processor in use on the existing hardware. Mixing radically different nodes into the same architecture is a critical benefit of VergeOS.
Protecting Against Server Failure
Using existing hardware enables a much faster return on your investment, and the flexibility of mixing hardware creates a platform that will run for decades. However, extending server life means a higher chance of component failure, so VergeOS provides advanced critical telemetry warnings of potential server failures. It also provides the ability to move workloads to another server within the data center seamlessly and with as little interruption as possible.
Extend the Life of Storage
Modern storage media should last at least 8 to 10 years. If used efficiently, most flash storage purchased in the past four years should meet an organization’s performance demands for over a decade and is only halfway through its usefulness. The need for more capacity can easily be met by extending the storage via a scale-out approach using less expensive QLC flash or hard disk drives.
Drive failure during the five to ten-year range remains relatively low and, with adequate redundancy and data protection, should pose no significant risk to the organization.
Why Do We Still Have Storage Refreshes?
Despite the potential longevity of storage media, most IT leaders still must plan for a storage refresh or replacement every four to five years. The need for storage refreshes is often vendor induced. They propose egregiously expensive out-year maintenance contracts to their customers to make them buy new hardware. Even so-called “evergreen” programs that refresh your storage every four to five years are subscription services that charge you for the upgrade in advance. As long as vendors continue to couple storage hardware and software, IT will always be at the mercy of these antics.
Attempts to converge storage via hyperconverged infrastructures (HCI) have failed mainly due to a lack of scale, performance, and high cost. Another challenge is HCI’s lack of node flexibility, which impacts both processor choice and storage media selection. In most cases, each node in the HCI cluster must have the same media configuration and capacity as the other nodes. As a result, most HCI clusters face the same storage refresh cycle as dedicated SANs or NAS solutions.
UCI Extends the Life of Storage
VergeOS provides a complete complement of storage services, including deduplication, drive-failure protection, and immutable snapshots. Storage media is housed inside the computing servers, eliminating the cost of standalone storage and its dedicated network architectures. VergeOS abstracts these services entirely from the storage hardware, and IT planners can use any media combination. IT can add media to existing nodes if bays are available. If not, they can add additional nodes. Just as VergeOS can mix nodes of different compute characteristics, it can mix nodes of different storage media types. Organizations will experience improved IOPS delivery, capacity efficiency, and reduced latency with VergeOS. Finally, The same redundancy that ensures seamless virtual machine failover also ensures continuous data access if a drive or entire node fails.
Extend the Life of the Network
Networks already have a relatively long life expectancy. Today’s data centers are deploying 25Gbps or greater Ethernet, which is sufficient for most applications. The next upgrade in bandwidth will be more because of part obsolescence than anything else. Some workloads need this higher bandwidth. IT needs the ability to add it incrementally and precisely.
Extending the network’s life is more about lowering costs, simplifying management, and, recently, dealing with product availability. Global supply-chain issues are impacting networking manufacturers more than any other aspect of IT hardware.
UCI Extends Network Flexibility
VergeOS’ role in networking is to bring greater flexibility and eliminate the dependence on proprietary hardware. While VergeOS can support existing networks, it includes complete functionality integrated into its common code base. It moves well beyond the “virtual switch” capabilities of HCI and hypervisors by providing complete layer one and layer two networking, firewall, DNS, and other related services. Controlling the network also means improved east-west traffic efficiencies enabling greater scale and seamless failover during node or site outages.
UCI’s integration of computing, storage, and networking into a common code base delivers the efficiencies that enable you to safely extend the life of IT infrastructure. VergeOS dramatically simplifies IT because there is now one component to manage instead of a dozen or more. This simplification manifests in daily provisioning tasks and eliminating dreaded weekly tasks, like “patch Tuesdays.”
To learn more, watch the webinar I did with VergeIO’s SE Director Aaron Reid, comparing VergeIO’s UCI to HCI vendors, Nutanix and VMware vSAN. We also take you through a demonstration of the features that make UCI different from HCI.
Leave a Reply