Adopting virtualization in the data center makes a lot of sense, especially if you are looking to optimize performance and reduce hardware costs. Virtualization offers a variety of benefits, particularly if you are hoping to increase resource availability and incorporate cloud capacity for applications such as big data. Of course, along with the benefits there are challenges as well.
One of the drivers of virtualization in the data center is the cost savings from using existing x86 server hardware for software upgrades. Infonetics Research also reports that 76 percent of the North American companies they surveyed see virtualization as an important driver for new security solutions. The survey also found that previously isolated virtual appliances are merging with software-defined networks (SDNs) to drive budgets up 57 percent in 2016. And research firm Technavio predicts that the global virtualization market will jump from $10 billion in 2014 to $21.5 billion by 2019, largely because of increasing demand for cloud-based data systems.
So what are the advantages and obstacles of adopting virtualization in the data center?
The Benefits of Data Center Virtualization
Virtualization abstracts the hardware components of the data center and provides a common software layer to manage the virtual machines (VMs). There are different types of virtualization for servers, storage, and networks. In the context of the data center, virtualization typically means configuring servers as VMs, in which each application has its own applications running in isolated partitions. When data center services are virtualized, there are no more islands of memory, storage, or networking, and VMs share computing resources to facilitate things such as load balancing or expanding processing capacity.
Data center virtualization offers a number of immediate benefits:
- Fewer servers – Virtualization allows you to use fewer machines in the data center, which reduces operating cost as well as problems such as heat buildup.
- Rapid deployment – Virtualization makes it easier to deploy new applications without hardware configuration. Backup and redeployment, for example, are easier because snapshots are taken throughout the day and can be migrated to new VMs as needed.
- Dedicated servers – Using virtualization, it’s easier to segregate services running on the same server, such as separating email, Web services, and the database server using VMs.
- Easier cloud migration – VMs can be deployed to and from your data center, thus making it easier to migrate to a cloud-based infrastructure as needed
Of course, even with all these benefits, data center managers have a number of concerns they must address when bringing virtualization to the data center.
Challenges with Virtualizing the Data Center
So what are the common problems with data center virtualization?
1. Missing components
One of the biggest problems is that IT organizations usually virtualize part of their data center assets. Like big data, virtualization works best when it includes everything and there aren’t silos of data storage or data management appliances. Limiting the scope of the virtual infrastructure ultimately adds cost and complexity.
2. Overcoming chaos
Over time, the proliferation of purpose-built devices that aren’t part of the virtual systems creates unnecessary data center complexity. Consider backup hardware. Companies invest in local and remote backup and disaster recovery including storage, backup servers, and deduplication appliances. These CPUs are idle when not in use if they are not running as part of a virtual data center. And there are probably redundant system backups so the same data is backed up repeatedly, thus wasting processing time and resources. Using virtualized systems eliminates underutilization and unnecessary redundancy, which can mean huge savings in the case of big data projects.
3. Underused servers
Most standalone servers have less than 10 percent utilization. Virtualization makes better use of existing servers while separating resources and workgroups.
4. Resource challenges
Allocating resources in the virtualized data center can create load-sharing problems. For example, where multiple workloads are multiplexed using the VM hypervisor, IO streams start competing for available resources. This increases the IOPS needed for virtual workloads. The usual solution is to overprovision the hardware to improve performance.
5. Managing portability
Virtual machines are portable, but there are constraints around that. VMs are tied to data stores within the virtual domain, and physical storage is often managed at the element level than as part of the virtual infrastructure. Policies are also configured at the element level and are not linked to VMs but to the storage elements. This means data center managers must manage all their data center resources using a top-down approach, establishing storage and policies at the VM and workload level to make them portable.
6. Realigning staff expertise
Ideally, virtualizing the data center eliminates the need for manual data center processes and much of the expertise behind them. Abstracting the hardware eliminates much of the hardware complexity, so data center staffers must realign their skills to manage applications at the virtual level rather than dealing with hardware configuration.
The benefits of adopting virtualization in the data center clearly outweigh the challenges, but it does require new thinking on the part of network architects. As IT professionals re-allocate network resources for new applications such as big data, understanding the limitations as well as the virtues of virtualization will prove invaluable.