Nutanix AOS 5 – greasing up the legacy apps
- Summary:
- A major revamp of its hyper-converged environment has come out of Nutanix, with one eye on AWS and the other on helping users of legacy business critical applications make the most of them in the new hardware world.
There is an on-going need for users to square the circle of getting new technology to play ball with the realities of established legacy applications. And - no matter how much they may be hated and a pain in the butt to work with – they are not going to go away any day soon.
The best that can be hoped for is that the new technology will contain sufficient tools to build a sensible bridge between those business critical legacy systems and the new opportunities it can offer. This is the nettle Nutanix looks to be grasping more firmly with the latest version, 5.0, of AOS.
Several of the new tools included directly target the need for better, deeper integration of legacy business applications into the hyperconverged world that Nutanix is offering. The company is also targeting those users looking more closely at a move to the cloud, and in particular moving some applications and tasks to Amazon’s AWS environment.
The flexibility offered by AWS in managing workloads is now a prime goal and pointed out in a blog posting on Version 5.0 by the company’s Director of Product Marketing, Prabu Rambadran, which gives a pretty comprehensive breakdown of the new developments. He wrote:
To deliver on the vision of Enterprise Cloud and offer public cloud-like services within the datacenter, it is important for us to provide infrastructure services similar to what the public cloud offers. Different workloads have different infrastructure needs and it is important to provide services that can be “turned-on” and “turned-off” based on applications needs – all without having to touch the underlying physical infrastructure. This is exactly what AWS does and that is what we are working towards with our Enterprise Cloud Platform as well.
Staying licensed
One of the bigger areas of potential conflict between legacy applications and hyper-converged environments is the question of licensing. This is where the traditional regime of `one application per server’ was the rule, which means that an application won’t run if it is not on its assigned, licensed hardware. With hyper-converged environments, however, an application may end up on any suitably configured physical server, creating an operational contradiction.
The Nutanix solution to this is an enhancement to the company’s own AHV hypervisor, AHV Affinity Rules. This is aimed at workloads such as Microsoft SQL or Oracle that are commonly assigned to specific nodes for licencing purposes. The Rules now allow virtual machines to be `pinned’ to a specific host or a set of hosts.
It also allows the reverse to set up, so that VM anti-affinity rules can be applied with the goal of ensuring that specific VMs are never on the same host together.
The company has also added greater operational flexibility, with a CPU/Memory Hot Add capability. This should prove particularly valuable to legacy applications users for it will allow administrators to add vCPU and memory to a running VM without any impact on service provision. As Rambadran wrote:
As applications evolve, it is important to dynamically adjust resources assigned to them so that there is no performance impact for the end user. This is what the hot add feature will enable. In the 5.0 release, this feature will be available in Tech preview and is expected to be generally available in a subsequent release. With these capabilities, AHV is ready for all your production workloads and there is very little reason as to why any virtualized workload cannot be run on Nutanix and on AHV.
There is also a development aimed at supporting business critical applications staying up and running when the workloads get heavier. This is Acropolis Dynamic Scheduling (ADS) which extends the existing intelligent initial placement capability so that the hypervisor can detect CPU, memory and storage controller hotspots and move deployed VMs to a better suited host:
Legacy hypervisor solutions factor in CPU and memory alone while making similar decisions. But with AHV, the VM placement algorithms will also factor in storage resources as well as storage controller bottlenecks before making a decision.
Other developments include Acropolis File Services (AFS) which are now generally available with the 5.0 release. This is pitched as a replacement for Network Attached Storage, which is now on the way out of a growing number of installations. This gives the ability to consolidate virtual machines and associated file data on the same cluster, working on ESXi and AHV hypervisors for a wide variety of use cases.
Acropolis Block Services (ABS) has also been enhanced with support for dynamic load balancing and flash pinning. This now offers CHAP for safer, more secure client-server conversations, as well as online resizing for higher availability. In addition, Oracle has now joined the list of certified workloads on ABS.
And something for the SMBs
The SMB user community has also received some specific attention in the 5.0 release with the provision of single-node backup/replication for native backup and data replication. This can be to a storage-heavy Nutanix node – which offers up to 40 Tb capacity - on the same site or a different, remote site.
The release also adds Just-in-time Capacity Forecasting for Prism Pro to provide users with modelling of infrastructure needs based on application requirements before those applications are deployed:
This way, IT not only understands application-centric capacity usage patterns and what can be done to optimize existing capacity, they can also plan their infrastructure needs for the future and deploy infrastructure if and only when they need them at `byte’ sized granularity all based on recommendations from Prism.
A separate blog, authored by Deepa Pottangadi, Instructional Designer, Nikhil Bhatia, Staff Engineer, Manoj Sudheendra, Member Technical Staff and Viswanathan Vaidyanathan, Member Technical Staff at Nutanix, is also available outlining the addition of Life Cycle Management (LCM). This is a new feature in AOS 5.0 that allows users to update the software and firmware of Nutanix clusters. LCM can be installed separately, and going forward, it will have its own release cycle.
This means that Nutanix will be able to deliver LCM updates more frequently, and it is expected to release a flurry of updates over the next few months. These will aim to provide support for a wider range of platforms and will exploit the fact that LCM can update itself.
My take
Interesting additions in this large update from Nutanix that fill out the support available for legacy business critical applications, which suggests that many of them will still be around for a good while yet.