Dell Technologies’ IT organization is accelerating its software-defined strategy to simplify and improve the compute and storage experience for users across its environment.
What began as an effort to incorporate software-defined storage into Dell’s data center has expanded to become the common compute and storage platform for the majority of the workloads within Dell’s modern data center to deliver flexibility and performance.
At the heart of that transformation is Dell PowerFlex, a software-defined, hyperconverged infrastructure platform that features compute and storage layers running on Dell servers, with networking components and software to simplify infrastructure management.
As Dell Digital, Dell’s IT organization, modernized its applications and migrated them to its data center over the past several years, we needed a way to modernize the accompanying databases to achieve that same flexibility and agility. We chose Dell PowerFlex, which supports everything from bare metal to virtualized and containerized infrastructure with one common footprint.
About a year ago, we finished converting some 65 petabytes of legacy databases to a new software-defined architecture hosted on 1,300 nodes of PowerFlex in our data center. We have since doubled the PowerFlex footprint to 2,600 nodes and more than 130 petabytes of installed capacity. And we expect to double that capacity again over the next year.
Achieving database agility
Evolving PowerFlex infrastructure has been a part of Dell’s strategy for its modern data center since 2016, as it merged two large, traditional data centers into its broader environment.
PowerFlex provides flexibility to run a uniform storage platform across all of Dell’s different types of workloads—including Oracle databases, MongoDB, Cassandra DB, Microsoft SQL Server, and Hadoop clusters—and create a unified experience. It provided a common framework to consolidate and modernize databases that were spread across the company and were using a variety of Dell products.
Our storage is now pooled, rather than being tied to different silos and arrays, maximizing storage capacity. Its massive scalability has improved database performance by at least 30 percent. (See case study Dell Digital Supercharges Performance with PowerFlex)
Initially deployed as a software solution on PowerEdge servers, PowerFlex has evolved into a fully engineered software-defined infrastructure with integrated IT operations and life-cycle management. PowerFlex is what we are now using for the vast majority of new storage and compute in our data center.
PowerFlex allows us to mix and match nodes within the same cluster, which is very flexible. So, we can have some nodes that have a certain profile and other nodes that have a different profile. We can also consume that infrastructure as storage only or compute only or both.
With PowerFlex’s simplicity and standardization, IT can offer database and application blueprints and designs to allow owners to easily provision new infrastructure for migration into our modern data center.
New options for flexibility
A key part of our efforts to increase flexibility, stability and performance in our data center while simplifying infrastructure management is creating a more agnostic environment so we have less specialized infrastructure.
We are using the increased capabilities of the PowerFlex platform to drive more abstraction between, for example, Windows VMs and SQL VMs, and other VMs so they can live together on the same cluster. We are also able to provide a more common setup for our VMware Tanzu Application Service (TAS) and VMware Tanzu Grid (TKG) environments.
In previous iterations of our environment, the need for specialized infrastructure added complexity. Cassandra required a certain architecture, our TAS and TKG environments required a different specific architecture, and our VMs required yet another architecture. PowerFlex is providing the foundation for an agnostic layer in a common compute and storage platform which allows us to provision workloads more like a public cloud.
PowerFlex also simplifies the onboarding of new functions in our environment. We don’t have to do rack reservations for specific cluster types in our data center in advance that may never be used. We can just deploy, build out everything, such as compute and storage and networking. Then when a user for instance needs another TAS cluster, we’ll just carve that up and boom, they get it. They don’t have to sit around and wait.
More new features, less down time
PowerFlex is letting us accommodate more workloads with less downtime for maintenance.
Among the features we’ve unveiled recently is a new GPU (Graphic Processing Unit) option for our Data Science as-a-Service within the platform. GPUs offer high-performance image processing through rapid mathematical calculations, offering high-performance analysis that data scientists require.
We also introduced a new high CPU, high memory option, so we’re able to offer double the compute and memory we had previously within our environment. That means we’re able to take on more and more use cases in our general-purpose infrastructure. In addition, we’re bringing in PowerFlex manager for monitoring and operational enhancements to the environment.
Faster patching and maintenance
We’ve seen massive growth in our internal IT systems that’s been driven with PowerFlex as the common compute and storage platform.
As our environment grows, one of the challenges is keeping all the various components up to date from the networking, storage and compute pieces. It is all integrated on one platform, but each component has different patching and maintenance regimes. We are introducing a capability called fault sets in our new PowerFlex design that will allow us to speed operational activities and patching.
Fault sets will allow us to deploy our new environments across eight racks and be able to take an entire rack down for maintenance without impacting the availability of storage or compute in the environment. That will significantly increase the velocity and parallelization of patching efforts within our environment.
From my perspective, I want to increase maintenance velocity, resiliency, and availability while decreasing the complexity of the infrastructure we are building. These are things that we’ve been working on for the past year and PowerFlex is the cornerstone of that.
We’re not tied to a specific way of consuming or deploying our environment. It’s very flexible. After all, it’s in the name.
For more information about PowerFlex or the recently introduced APEX Custom Solution: PowerFlex for Databases – which includes flexible payment options and Dell Tech Managed Services so you can focus on your databases while Dell manages the infrastructure – please visit the following resources:
