olly - Fotolia
Virtualization trends in 2018 portend the arrival of a number of formerly edge technologies that will continue to reduce the future of virtualization technology from a product to a feature.
The ever-evolving nature of technology means that IT admins need to adapt to survive. In the shadow of the cloud, virtualization administrators need to examine developing technologies and evaluate their advantages as they become mainstream.
Serverless computing makes the cloud accessible
Despite its name, serverless computing doesn't actually function without servers. Serverless is basically a cloud-based scripting engine that enables easy access to various cloud-based services.
To make a serverless function, you must define a trigger that will activate a serverless script. This could be a file being uploaded to Amazon Web Services' Simple Storage Service, for example, or a message delivered via a message queuing system.
Triggers make serverless more powerful than a batch of files for the cloud. It's like a more capable version of CRON, if CRON ran on a server on which you could run images through advanced compute features.
Big data analysis tools
Machine learning, artificial intelligence, business intelligence, facial and voice recognition, and other forms of advanced analytics are the new normal.
Ten years ago, we were just starting to come to terms with the concept of big data, and we struggled to collect, store and analyze it. Now, we have storage technology designed to serve as high volume data warehouses -- these come as object storage, NoSQL stores and next-generation relational databases. Complicated, black box algorithms that work on the data have replaced more simplistic analytical tools.
Multi-cloud enables big data mobility
Multi-cloud platforms will emerge as one of the top virtualization trends in 2018 because they address the problems public cloud providers have with cost and interoperability when using data across providers. Multi-cloud services come in three basic forms.
Multi-cloud management tools are the first type. These don't actually solve the problem of using data sets in multiple clouds, but they make moving them around easier.
Centralized storage platforms are the second kind, and NetApp's Private Storage for Cloud is the canonical example. NetApp puts a bunch of their files in a third-party data center and offers that storage to public clouds via high-speed, fiber optic connections. You store your data in one place, which means you don't need to move data in and out every time you have to move between public clouds.
Multi-cloud storage fabric platforms are the third category. Fabric storage platforms move the data around the fabric to make sure that the data is close -- in terms of latency -- to the workload, wherever that workload might be executing. Multi-cloud storage fabrics are more complex than centralized offerings, but they are also more flexible.
Edge computing brings data processing closer
Not all data can be hauled to the central data centers of the public cloud. Privacy, data sovereignty, latency requirements and sheer volume might require the processing of data closer to where it's generated.
Public cloud providers aren't ready to enable their analytics tools to be used on premises, so edge computing is emerging to address this. Edge computing is basically a subset of public cloud services provided, in terms of geography and latency, close to where processing takes place.
Driverless cars are a prime example. If all driverless cars in a given location were to send their sensor data to a centralized processing facility, the data analysis tools available there could give all of the cars on that grid the ability to see around corners. All the cars would know what the objects in the area were doing anywhere that a connected car's sensors could see, making the entire grid safer.
For cars to meaningfully make use of this functionality, the latency would have to be far lower than what can be achieved by shipping the data back to centralized public clouds, processing it and returning it.
Configuration management scales administration
None of these 2018 virtualization trends are possible at scale if system administrators need to manually manage all of their workloads. To achieve this scale, administrators need to move toward composability and automation.
Configuration management is the first step in this direction. Configuration management enables administrators to define the configuration of Operating System Environments (OSEs), and even some application as code. An agent is installed in the OSE, the configuration is read by the agent, and the agent applies the configuration to the OSE and some applications.
If the OSE or application's configuration is changed, the configuration management agent can detect the change, reset the configuration back to what was defined in the code and generate an alert to the administrator.
Infrastructure automation and containers threaten virtualization
The infrastructure that lives under the OSEs calls for automation, too.
Every major infrastructure provider -- both on premises and in the public cloud -- has its own proprietary platform. VMware administrators are familiar with vRealize Automation, for example, while Amazon Web Services administrators are familiar with CloudFormation.
For cross-provider infrastructure automation, Terraform is emerging as the standard, much as Kubernetes has largely led the container orchestration wars. Combined, Terraform and Kubernetes provide a powerful infrastructure automation platform. There are few more impactful virtualization trends in 2018. This technology could make containers serious competition to VMs and portend disruption to virtualization technology as a whole.
Composable infrastructure and infrastructure as code
Composable infrastructure consists of infrastructure automation and configuration management tools, along with things like scripts and application configurations that enable the separation of a workload's data from the infrastructure and the configuration of that infrastructure.
Composable infrastructure, infrastructure automation and configuration management are all in the infrastructure as code space.
Configuration management services are increasingly trying to automate infrastructure other than OSEs, and infrastructure automation tools are increasingly moving into the configuration management space. In a perfect world, composable infrastructure would automate everything except for racking and connecting equipment.
Virtualization administrators must adapt
Virtualization trends in 2018 won't revolve around virtualization. All of those features built into vSphere and System Center Virtual Machine Manager that we've spent our careers mastering ultimately amount to line items forming sub-features of a single feature of the cloud. And clouds -- plural -- are just one component among many in today's ever-changing data centers.