WavebreakmediaMicro - Fotolia

Why enterprises should dump their VMs for containers

Kubernetes is five years old and server virtualisation is the new legacy, says Aparna Sinha, director of product management for Kubernetes at Google

Container technology has reached a stage in maturity where enterprises now have a stable platform on which to base a next-generation agile IT architecture. During the Kubecon-CloudNativeCon joint event in Barcelona, the Kubernetes container orchestration platform celebrated its fifth anniversary.

Aparna Sinha, director of product management for Kubernetes at Google, says enterprises have much to gain by migrating workloads from virtual machines (VMs) onto containers.

Sinha was at Google at the very start of Kubernetes and now runs the Google Kubernetes Engine (GKE) managed service offering. “I have been with Google for six years, and with Kubernetes for four years and have seen it grow up from a baby,” she says. “Thanks to the managed service, we get direct interaction with users.”

Large enterprises are now adopting Kubernetes at scale, and Sinha says Google is often asked what software stack should be used by enterprise developers. “I am certainly seeing the adoption of Kubernetes at scale,” she says. “It is top of mind.”

Depending on the size of the enterprise, in her experience the CIO is taking notice. “I am in discussion with large banks, larger insurers, all of the retailers, healthcare and I am starting to see interest in Kubernetes in heavy industries too,” she says.

Kubernetes provides a means to manage and orchestrate containers in an automated way. It provides automation that enables enterprise developers to build applications for the cloud era.

Traditional industries require a driver for change, which is why mainframes have not gone away. For Sinha, VM technology is becoming the new legacy IT. “The reason why people adopted virtualisation was cost reduction, to get rid of their reliance on hardware,” she says. “You could pack a lot of virtual machines into the same piece of hardware.”

The industry talks a lot about cloud native, which allows  IT infrastructure to scale up and down elastically and enables workloads to be moved without downtime.

“When I ask users why they adopt Kubernetes, they say their applications become more reliable”

Aparna Sinha, Google

But according to Sinha, virtualising servers does not necessarily lead to a cloud-native computing architecture. “With virtualisation, you are not really scaling based on the application,” she says. “A lot of scaling is manual.”

Sinha claims the use of containers leads to a much greater level of efficiency than with server virtualisation. “When I ask users why they adopt Kubernetes, they say their applications become more reliable,” she says. “When your machine goes down, Kubernetes reschedules the application to run on a machine that is up.”

It also auto-scales up and down, which, according to Sinha, has made GKE the preferred platform for hosting e-commerce sites.

Turning to retail, she says: “We had a very successful Black Friday on GKE. We are known in the industry as the place to run your e-commerce site.”

The fact that the system scales up and down seamlessly means IT departments do not have to over-provision capacity, says Sinha.

“Enterprises are always looking for cost savings, so you don’t have to provision for peak [usage],” she adds. “Even if you have no idea what your demand will be, the system will scale up and down. It also enables multiple applications to run on the same system.”

Sinha claims that GKE enables these multiple workloads to be compacted. “Some people have told me they moved from VMs on AWS [Amazon Web Services] to containers on GKE and they got up to a 90% improvement in utilisation,” she says.

Developer productivity and customer engagement

For Sinha, the third benefit of containerisation is that it gives developers a way to change their applications hundreds of times a day, which is important when companies want to improve how they respond to customers. “If you can get scalability, lower costs and improve developer productivity, why stay in the old world?” she asks.

Sinha urges traditional businesses to find ways to free up IT budget by moving workloads to containers in the cloud. “You want to be prudent and do more with your budget,” she says. “Use the cloud. It offers such great efficiency, which can help you save money to invest in IT modernisation, using smart data to serve your customers better.”

In Sinha’s experience, a lot of large companies are sitting on troves of data, which they are not using fully. “When they do start using this data, they often find they can increase customer satisfaction, personalise and serve customers in a much more meaningful way, cut costs or improve the efficiency of their supply chain,” she says.

This leads to a new relationship with customers. As an example, she describes how a traditional car insurer could use traffic data to analyse accident blackspots and position recovery vehicles in those locations. This means they are likely to be close to a traffic accident should one occur, so their customers’ vehicles will be recovered from the accident three times faster.

Read more about Kubernetes

  • Online betting company Bet365 has begun using Kubernetes to make it easier for its IT infrastructure and product teams to share complex configurations.
  • Price comparison site MoneySupermarket.com has used container orchestration to split its machine learning data pipeline into parallel processes.

“Nobody tells the customer we are using machine learning and GKE, but the user sees the benefits,” says Sinha. “These days, this is really important. If you don’t do it, there will be a startup that takes public data and develops its own service. If the traditional enterprise does not modernise, it will become obsolete.”

What excites Sinha as a Google employee is that technologies such as machine learning, personalisation, intelligence, context-aware technology and use of external data are coming from Google. “If you marry this tech with the enterprise, it creates great power,” she says.

Kubernetes was derived from technology that Google developed internally to manage its web scale infrastructure, and it obviously did something right to scale globally. Now this technology is available in a form that is ready for enterprise IT. 

As organisations embark on IT-powered strategies to improve efficiency or boost customer engagement, Sinha believes now is the best time for IT to start taking Kubernetes and containerisation seriously.

Read more on Containers

Data Center
Data Management