A decade ago when cloud providers like Amazon entered the market, they redefined how to provision IT services. These services provided self-service capability that raised the bar for how users and application owners wanted to interact with IT. The mission was set; IT needed to become more cloud-like and modernize their data centers. Concepts like virtualization, containers and software defined networking empowered organizations to achieve the “cloud-like” goal. Storage, however, is the boat anchor holding the data center back from its modernized future.
The Storage Problem
A major problem caused by storage is its lack of multi-workload support. Data centers, more so than ever, have a variety of workloads ranging from traditional bare metal database applications, to virtualized and containerized applications to vast amounts of unstructured data. Within each of these workload types there are also varying performance and capacity demands.
Data centers also face a location problem when it comes to storage. Most organizations today have multiple data centers, dozens of branch offices and at least some cloud based data. In most cases, the organization wants to place data as close to the user or application as possible.
The legacy solution to the multi-workload problem is to implement a separate storage system for each workload type, making automated provisioning of storage resources, when deploying new applications or workloads, almost impossible. Forcing the organization to purchase a separate storage system for each workload also adds to the overall cost of the infrastructure and makes operational excellence unattainable.
Traditional scale-out systems claimed to address the performance and capacity issues but because each scale-out system focused on a particular workload type, the data center still ended up with individual silos of scale-out storage systems. In addition, most on-premises scale-out storage systems are very rigid. While expansion can be made by “just adding a node”, those nodes typically must be very similar to other nodes in the scale-out cluster and those clusters provide weak support for mixed media types.
Neither legacy storage solutions nor scale-out storage solutions can address the multi-location problem. Each requires a separate silo of storage per location and again per workload type at each location.
Datera is a software defined storage solution designed to support multiple workload types in a single cluster and to span multiple data centers, including the cloud. The solution’s design delivers the performance and protocols that legacy applications need while scaling to meet the demands of more modern environments. Datera’s goal is that the solution scale to meet an organization’s workload requirements no matter where that workload is best suited to run.
The Datera solution is a scale-out architecture designed to span multiple organizational campuses and even into the cloud. It effectively creates a data fabric that spans all the locations that the organization has. The solution, the Datera Data Services Platform (DSP), uses replication instead of erasure coding to protect data. The use of replication enables the storage cluster to service all reads for an application from a single node, reducing the network latency created by erasure coding. Datera’s solution can also effectively manage different types of media, placing more active data on flash storage and less active data on less expensive high capacity hard disk drives. The combination of these two features allows the solution to meet production performance requirements and the cost-effective requirements for long-term data retention.
The DSP is also location aware. Policies can be set to manage data distribution between data center campuses, remote locations and the cloud. Once data is in the cloud, cloud compute resources can operate on the data for a variety of purposes.
IT professionals are commonly presented with environment specific and workload specific types of storage solutions. The problem is that each of these solutions creates its own island of storage, which the organization must independently purchase, manage and protect. The design of Datera’s Data Services Platform solution simplifies IT storage architectures, creating a single design that spans workloads, locations and use cases. For organization’s drowning in islands of storage, it deserves strong consideration.