For IT teams managing enterprise business applications, Total Cost of Ownership (TCO) is critical. Data growth requires additional resources, but adding new hardware is not always viable. Enter virtualization. As its popularity grows, virtualization can offer lower initial costs, ease of scaling and centralized resource management. However, virtualization comes with its own set of challenges that are worth exploring.
Virtualization Challenges to Consider
Enterprise-level hardware can provide a large number of processors and amounts of memory. Current hardware specifications have all but eliminated hypervisor limitations, and hypervisor technologies offer unprecedented memory and CPU management, high availability and fault tolerance.
Organizations should keep in mind, however, that virtualization does come with some challenges:
Virtual machine (VM) sprawl and management: For years, we have watched data center server sprawl taking up valuable resources and capital out of IT budgets. Caution is still necessary because it only takes a few minutes to create a new VM on the server. You need to have proper policies and procedures to help control server and VM growth.
Security: Though no system can be guaranteed to have 100 percent protection, technology is available to prevent and manage attacks. To ensure the best levels of protection, hosts and VMs need to be properly configured.
When Virtualization May Not Be Best
There are also some use cases where virtualization may not be the best solution:
Machine learning: Modern AI is computationally intensive, making TCO a crucial decision-making variable. Developers prefer GPUs for training models, as they can access far more cores in graphics cards. But virtualization has been making leaps to get closer to baremetal efficiency. Not all hypervisor and operating system vendors are created equal. VMware and Windows are clearly market leaders. You should carefully analyze the organization's requirements before selecting the right infrastructure.
Media and gaming: Performance is critical for video and gaming. Yet, streaming media can quickly consume CPU, RAM and bandwidth resources. Another possible issue is latency. Low latency is vital for a fluid gaming environment, while sensitive media, such as medical imagery require compliant storage. This generally leaves private/hybrid clouds or bare metal as the only viable options. Many organizations further reduce latency by utilizing multiple data centers.
The trend toward virtualization will continue, as it offers low CAPEX, thus a lower barrier of entry for high performance computing. But as organizations grow, their IT needs can skyrocket, thus bringing OPEX into management's focus. Perhaps this is why Gartner says that the server virtualization market is reaching its peak, but Market Research Future recently forecast that the market will continue to grow at the rate of 7 percent CAGR between 2017 and 2023.
Whether virtualization market has peaked or not, it's definitely here to stay. So while the temptation is to jump on the latest trend, we in IT already know the price of blindly jumping into a new technology without due diligence.