As director of Dell Technologies’ modern data-centre division, Amit Mehta looks after the needs of enterprises that are wanting to move from on-premises to off-premises, as well as willing to adopt multi-cloud environments for meeting ever-increasing storage and compute demands.
In a talk with TechCircle, Mehta sheds light on how public cloud adoption has increased owing to a number of reasons, why enterprises are moving towards multi-cloud environments and how Dell is keeping up with the demand for increased storage requirements. Edited excerpts:
How has the outlook of enterprises towards public cloud changed?
I have been following this market since 2011, and the catalogue of services that companies provide has grown. Mission-critical providers such as Oracle or SAP and a bunch of other providers have solutions in the public cloud. Today there is data sovereignty and most data centres are situated in India. The government itself has adopted the cloud in a big way so there is a more positive outlook towards the public cloud.
When an enterprise calls for a requirement, the first step is to see whether that service can be fulfilled through SaaS (Software as a Service). Dell uses Workday for all our HR applications, which previously used to reside on-premises. We also use Salesforce for our management of the sales funnel and Office 365 from Azure. These workloads can easily be serviced on a SaaS model.
Is there a move towards self-service-driven models and infrastructure?
The shift that is noticeable now is that the earlier model used to have less build and less construct, and it used to be a service-driven mechanism. The time to value in the older models was very less and it is considerably better today.
The IT organisations now use models based on IaaS and PaaS (Infrastructure as a Service and Platform as a Service), which are self-service-driven, and are all under an automated and orchestrated infrastructure, along with a bunch of converge infrastructure that helps with the same. All of these trends are fuelling the move to the public cloud as well as hybrid cloud strategies.
Are you moving towards more SaaS models within Dell to help clients better?
We have a case study, when Dell and EMC came together to build infrastructure for thousands of employees. The question to answer was how aggressively we are going to virtualise our data centres. We want to go towards 90% virtualisation environment.
We had a traditional waterfall model of developing applications. So we ate our own food and applied Pivotal (a partner of Dell); Pivotal helped us convert the monolithic model into a microservices architecture, and now we are able to roll out applications on the fly.
For example, on a black Friday sale, we were requested to build, test and deploy an app in a couple of days. In the earlier world, it wasn’t possible to turn this around. But with agile methodologies, we were able to deliver the app by Saturday, test it by Sunday and make it live on Monday.
The whole agile development methodology based on a microservices architecture was the reason behind being able to build an app within a couple of days, which is exactly the culture that we want to take to our customers. We use the multi-cloud ourselves; we are moving our SAP to a virtual stream cloud.
What are the current demands of enterprises when it comes to data storage?
The sheer number of data sets being generated is huge. What is driving the growth is the system of records, but the system of engagement has also increased and there is a spurt in unstructured data owing to the higher number of devices we use today.
From the consumer standpoint, it is about how we interact from mobile devices and corporations, how we can have a system of records and what needs to be done from a compliance standpoint for that data set.
From an enterprise employee perspective, we work across different domains. IoT (Internet of Things) and sensors are churning out a lot of data. Edge computing, data centre computing at the core and cloud computing have fuelled the growth. This has put pressure on the storage and on the compute systems.
How are the vendors keeping up with the need for increased computing power?
In the compute space, there is a lot of push for insights and value. With Hadoop, you can intersect data sets in very meaningful ways. Hadoop today has leveraged data to provide a 360 degree customer experience management, real-time campaign management and other related solutions to CIOs (chief information officers).
The computational power requirement is huge. This is complemented by the move towards GPU (graphics processing unit) computing when it comes to massive data requirements for AI (artificial intelligence) and machine learning. Today enterprises are asking for a GPU model that can complement their AI/machine learning requirements.
Just the number of cores and sockets that are being packaged in a small footprint is tremendous. In the age of IoT, you have a lot of Big Data as well as Fast Data that comes through.
How has the storage need changed? What are the new computing strategies and how are they classified?
There is also a move towards scale-out storage platforms to take in data that is coming from multiple sources. In such cases, you cannot have traditional environments scale up architectures and storage.
The data storage also requires higher density; some of our devices can hold 54 petabytes of data. You can manage 54 petabytes today, the way organisations previously managed 54 terabytes. That is the level of simplification and management capabilities that has come into the system.
Our flash-based storage platforms can give microsecond storage latency to respond to customers.
As the data sets grow in size and scale, you start to deploy storage and compute strategies based on workloads. Data is also treated based on whether it is performance-centric or capacity-centric.
Which verticals are traditionally seeing a lot of traction for the multi-cloud? What are the upcoming industries for their adoption?
There is a growing multi-cloud trend adoption in many verticals, especially in the IT vertical, where they will deploy Office 365 in Azure, Oracle in an Oracle cloud, and SAP system in a SAP service provider.
For mission-critical applications, the infrastructure can be stored on-premises, and they will have an operations control that will manage data across all these facets.
Other than IT, the government and the unicorns born and matured in the cloud are utilising the multi-cloud in a big way.
How can Dell’s storage platform help with newer use cases for storage?
Our Dell Technologies’ platform gives consistent infrastructure that can be managed by an on-premises and off-premises environment, irrespective of whether it is VMware-centric, or Kubernetes-container-centric. One control panel can help manage across different clouds.
Other use cases include video intelligence and surveillance based on AI; all the cars today are being equipped with a variety of apps that are data-centric.
For AI specifically, customer experience management and robotic process automation are areas where storage is going to be used in a big way.