In the proverbial quest to align the right price/performance with the value of the data, IT ends up with storage silos. This creates significant operational challenges and lock-in scenarios from multiple storage vendors. So one option is… store everything on flash! Makes sense right?
Yes, except that it is too expensive, and might not be aligned with the economic value of the data in those flash devices.
Datera answer is smart software that runs on standard x86 servers at performance and latency of enterprise-class storage! Through its policy driven and smart placement architecture, Datera Elastic Data Fabric enables customers an easy way to program the right storage technology to be aligned for the value of the data. And as the value of data changes over time, customers can easily move workload transparently across different storage platforms through simple policy changes. The end result? Unmatched operational efficiency with high performance, low latency solution that drives the absolute lowest TCO! And no hardware lock-ins!
Peter Drucker and Edwards Deming established the saying “You can’t manage what you don’t measure”. So true in today’s data center explosion. A decade back, it was the runaway costs due to the virtual machine crawl, more recently, it is about the unaccounted and unmanaged application growth on the public cloud. In all of these trends, one of the anchors has been unprecedented growth in storage costs. Enterprise applications requiring high-performance storage.
However, the lower tiers of storage have been the enabler for big data. For example, data dumps such as infrastructure logs and telemetry are now mined for insightful information. The reason is the economic value from this data is higher than the storage costs to house these data dumps. Can I say this has been one of the reasons for the recent explosion of digital data?
Now that we have established that there is a wide spectrum of applications leveraging different storage tiers, modern clouds really need storage-as-a-service rather than just many storage arrays. The right storage tier for the appropriate application, not just performance optimized, but also price optimized. Data is currency
Let’s establish the two meta points: (1) Not all data is equal (2) Different data has different economic value.
Modern clouds have a wide range of applications. No silo is safe. Test and development workloads, Hadoop and analytical workloads, high-performance applications and caching applications. Historically, performance and capacity were the metrics used to map these applications to the “right” storage tier. Performance includes IOPS, latency, working set ratio, and so on. A vector in performance is the notion of working set ratio (ratio of working set data to complete dataset). This ratio determines if all-flash storage or hybrid storage is appropriate for a given application.
One of the core value propositions of Datera storage system is “Smart-Placement.” Smart-placement is about understanding the application’s service level objectives (SLOs) and placing the data in the appropriate economic data tier. This placement can be explicitly defined, or let the system implicitly execute. The right storage tier for the appropriate application, not just performance-optimized, but also price-optimized. Data is currency.
With the Datera system, one can start with hybrid flash nodes and can add to that cluster. Hybrid flash nodes have an IOPS vs capacity gradient. For most environments, which require high performance but have a small working set ratio, this configuration is idle fit.
Another option may be starting the Datera system with all-flash nodes and adding to that cluster to scale. All-flash nodes have a different IOPS vs capacity gradient. While this configuration would work for all environments, there is the cost associated with this deployment (relative to the hybrid flash configuration). All-flash will be especially idle for applications demanding high working set ratio.
Well, clouds really need both and a number of variations in between. The Datera system uniquely offers the ability combine hybrid flash nodes and all-flash nodes today, with more node types in the future. This provides the wide price-performance aperture that clouds really need.
Now, the question is — so what?
For a service provider or enterprise customer offering IaaS or PaaS to its customers, this dramatically addresses the acquisition and operations costs.
First, let me address the acquisition cost aspect. Datera Smart-Placement allows for explicit placement of a volume as all replicas on all-flash, primary replica on all-flash nodes (and subsequent replicas on hybrid nodes) or all replicas on hybrid nodes. This allows clouds to not only provide multiple service tiers but also have different acquisition costs for each of these tiers. The right storage tier for the appropriate application, not just performance optimized, but also price optimized. Data is currency.
Now to the operations cost savings. Traditionally, cloud builders had a large monolithic storage array throttled to provide different services or at best have different monolithic storage arrays for each project or storage tier. Each having a unique operations model and having different utilization levels, painfully handcrafted to address each specific application. Any change in the scale or performance of the application breaks these deployments. This has been a key reason why many of the legacy clouds have failed. Clouds need scalability and agility.
Datera provides storage-as-a-service under one single control plane. Think of having multiple different storage arrays under a single plane, without the overhead of additional management software or operational overhead of different storage arrays. With Datera, a service tier can be configured based on volume placement (all-flash, primary flash replica, hybrid), IOPS, bandwidth, replication factor (IO durability), and latency). The right storage tier for the appropriate application, not just performance optimized, but also price optimized. Data is currency.
We welcome you to experience the Datera difference for your cloud deployments. Please contact firstname.lastname@example.org for a proof of concept, demo or additional information.