HPC Review interviewed Sudarshan Srinivasan, Director of Product Marketing who tells us more about the improvements of the new version of Nutanix’s operating system.

HPC Review: What are the key enhancements in this new version?

Sudarshan Srinivasan: We continue our strategy to make the underlying technical infrastructure invisible for administrators. We look further than hyperconvergence, an area in which we were pioneers, with the aim to provide companies with an integrated cloud solution.

HPCR: You put forward a performance improvement factor by 4. How did you achieve this result?

SS: The first area for improvement was to optimize the overhead of context switching between virtual machines. We also reduced file locking management overhead at the input-output stage and improved the checksum calculation performance. The other axis of major improvements focused on caching. So far, we used a single opload cache, we now use multiple opload caches, distributed both in memory and on all SSD integrated into our appliances in order to significantly improve performance. We also took advantage of the positive developments of version 11 of C ++.

HPCR: How does this performance gain translate into real usage and how can current users benefit?
SS: the performance gain at the application level is considerable and pushes the operational limits. This is allowing us to manage up to 30,000 Exchange mailboxes on a single appliance. To take advantage of all improvements to the system, it is sufficient for our current clients to update their operating system, it is completely free.

HPCR: Can you detail the new features of your app to app fabric?

SS: Our app to app mobility fabric is designed to simplify the movement of applications and data from the Nutanix hypervisor to a public cloud. Because we believe that the company should retain the choice of where its applications reside based on criteria like resilience or criticality. Our fabric is based on what we have called Nutanix Foundation, which integrates three major mechanisms.

The first mechanism does not require the physical movement of the application to be useable on a Nutanix cluster through a conversion performed by the Acropolis Nutanix hypervisor. This conversion is automatic, reducing the risk of manual conversion. The second mechanism now allows for backups between two appliances, even if they run different hypervisors. The benefit is being able to integrate in a heterogeneous technical environment. Finally, the third mechanism provides the company a cross-hypervisor disaster recovery feature while keeping the virtual machines in their native environment. A programmable snapshot replication feature makes snapshots at the disk level to other Nutanix appliances with a programmable retention policy. Snapshots take into account the VM data and the VM themselves. HSI VMs are instantiated according to the original.

The benefit of these new functions is to help the company keep the choice of the place of storage and execution of its virtualized environments, by allowing these operations on production environments, thus saving time and reducing the risk traditionally associated with any move operation workloads. In addition, it avoids creating an additional silo.

HPCR: You put forward artificial intelligence in this new version. How does it work?

SS: Our integrated management console has machine learning algorithms to automate as much as possible until then strictly manual operations. Thereby reduced most administration tasks to a single click, or even zero click if the administrator wishes to do so. Our predictive algorithm, or rather our predictive algorithms, as there are several of them, constantly examine the use of resources to establish a table of predictive load. This helps to meet the vast majority of current administrative operations (but not all). The special feature of our algorithms is to take into account changing uses over time to adapt. The addition of new predictive algorithms will be possible through a simple software update.

HPCR: You claim the right to make the IT infrastructure of the company “invisible” How ?

SS: We took advantage of this update to improve how administrators interact with our appliances. We have integrated an innovative search engine that allows them to interact directly with the desired functions. For instance, finding how to create a VM will guide them throughout the whole process on real resources and settings, and not limited to detailing them what to do. This results in greater reliability and speed in usual administrative actions. The second aspect on which we worked is the highly customizable administration dashboard, which summarizes the state of application resources, storage and network as well as their performance and even their predicted life span that takes into account the real use history. The administrator will thus can know at a glance if he has 60 days of storage, for example, to consider corrective actions, either recovering unused resources or adding resources to ensure uninterrupted operation.