The use of cloud-delivered applications and services gets ever closer to dominance amongst typical IT users and in the process continues to unveil new levels of functionality that offer both enhanced potential, but also increased levels of complexity that require new levels of management – or at least new levels of forethought and planning. And many of those new functionality levels have their roots in one word – continuous.
There is now continuous development of both new and additional/modified/updated applications code. The days of the three year applications update cycle are becoming historical curios, replaced as part of another continuous – in this case, code delivery. Daily updates and tales of leading edge applications providers running 300+ updates a day are no longer uncommon. Real time, continuous use of those applications, plus the monitoring and management of the events that follow as a consequence are now taken for granted.
Those applications are now packaged up in containers together with the resource configuration instructions needed to set up an instance and run it. The orchestration of containers, so that full business processes can be pulled together from multiple containers , is now possible with Kubernetes. This is already close to marking the beginning of the end for ‘applications’ as they have been known and loved/cursed for many decades, as the functionality that goes to make up a business process is broken into increasingly granular micro-services, another continuous joins the growing list.
This is continuous optimisation. With every micro-service having its own optimisation requirements, and each of them being swapped in and out of operation as workloads require, the days of one application per server and a three year gap between optimisations have become the stuff of legend. This is putting additional pressure onto many of the systems vendors, especially those involved in the storage, management and delivery of data, such as NetApp.
The company recently had its annual conference and as part of that held a press conference to talk through some of these changes and how the company is adapting to meet the evolving needs of users as they transition to new cloud-based services.
Continuously more continuous
The continuous optimisation piece is part of NetApp’s move to providing users with a data fabric which is intended to take the best capabilities from both the enterprise data center and the public cloud environments and ensuring they are each available to the other camp, as the company’s Chief Technology Evangelist, Matt Watts, made clear:
We look at the data fabric as two things. It brings the simplicity and flexibility of the cloud to the data center and it is also about taking optimisation and enterprise data services to the public cloud as well. It gives people the ability to feel more confident and more secure in terms of running some of their mission-critical applications inside the big public cloud providers.
So when it comes to that combination, he now likes to talk in terms of delivering application driven infrastructures. The goal here is to help companies get more value from cloud and be able to use it more with less cost. This is where the continuous optimisation plays an important role., he says:
Users need to be able to re-balance, to re-allocate the way that they're using cloud resources. That is much more aligned to the way that the application is actually running, what the demands of the application are at the time.
The focus of this is on three areas - storage, Infrastructure as a Service, and optimisation both for data and compute capabilities.
NetApp has, of course, been providing storage optimisation for a good while, but now it has reached the point where it can introduce optimisation services for compute as well, based on Spot. NetApp acquired Spot earlier this year for its compute management and cost optimisation capabilities on public clouds and is now using it to give a real focus on helping users reduce the amount spent on the compute costs of their overall environments.
The company now has storage services available on the three big gorillas of public cloud, with its Cloud Volumes ONTAP service providing data management capabilities for file and block workloads on Amazon, Microsoft and Google. This gives the ability to run a virtual version of its on-premises storage device in pretty much every region of every major cloud provider.
Watts suggested it is best to think of this as Volume as a Service. It is a virtual version of NetApp storage technology running inside one of the cloud providers and the key thing is users don't need to or want to manage anything that sits behind it, they just want a volume with characteristics and capabilities, a set of performance, and capacity:
What we're adding to that is the ability to begin to learn the different levels of compute inside the hyperscale providers, such that we can start providing a level of ongoing optimisation on top of what we've done for volumes, to be able to deliver that for the compute side of things as well.
Not just serverless, also storageless
A key driving force in all this, according to Amiram Shachar, Vice President and General Manager for Spot within NetApp, was the huge migration of customers onto containers, which created the ability to break applications down into smaller pieces, scale them independently and deploy them as micro-services, effectively the new paradigm of ‘function’.
Such functions then become serverless, as they are tied to a specific server, which does mean they have one downside - the need to be managed. This, he observes, is something users are trying to get beyond as they don't want to manage their compute anymore, just to ship independent pieces of code. Yet they want that compute and other services to be fully managed.
Meeting this need lies behind the company now introducing a serverless and storageless environment for containers, he says:
The main difference is that you don't have to manage them anymore. With serverless, you're getting compute that you don't need to manage anymore and with storageless we are providing storage that they don't need to manage anymore. Really, the time has come for applications to define the compute and storage they need. As they scale. We want just the applications to define what they need.
This, he suggested, is going back to application driven infrastructure, which will now have the Spot compute product integrated in to it, together with a new interface called Spot Storage. Working through Kubernetes, this interface is allows customers to get storage for their containers with no management whatsoever, Shachar claims. This will be part of Project Astra, an application data lifecycle management service for containers made for hybrid cloud environments. This will also include an advanced monitoring product for Kubernetes as part of its Cloud Inside Platform, he adds:
We've been working with customers for the past few months on defining exactly what that's going to look like. Usually in containers, infrastructure, customers are running their own control plane that's usually managed with their cloud providers, such as Elastic Kubernetes with AWS. What we've developed is an overlay of management that can connect with each one of these control planes so customers can choose their cloud, even on premises. Then we provide them automated, optimised compute and storage environments where they don't have to manage servers and storage anymore. This lets customers use their deployment of choice.
Yet more evidence that a significant long term change is under way in IT systems management, as the monster of rapidly increasing technological complexity is contained, tamed and trained to be ever more useful by automated management systems. It will be interesting to see what the next few years bring as this change continues.