Nimble Storage Takes A Single Architecture Approach To The Flash-Or-Disk-Storage Question
Nimble Storage recently announced a range of new capabilities for its Adaptive Flash platform. In addition to software-based encryption and a REST API-based extensibility framework, the platform now can deliver all-flash, auto flash, and disk-only service levels. Sean Roth, senior product marketing manager at Nimble Storage, explains the benefits of the platform’s single architecture approach and the benefits this has for you and your clients.
BSM: Can you explain the single architecture approach?
Roth: Today’s enterprise data center is tasked with supporting a wide variety of critical applications, and each of those comes with its own distinct set of service level agreements (SLAs). From a storage perspective, this means the storage infrastructure must be able to fulfill a broad spectrum of performance (input/output operations per second [IOPS], latency) and capacity requirements, while delivering an adequate level of protection and security for the data.
Organizations traditionally deploy multiple silos of storage from various storage vendors. This is because the vast majority of storage solutions can feasibly support only a subset of critical applications deployed by enterprises. These individual storage solutions tend to deliver on either high-performance/low-latency OR high capacity, but rarely both within the same platform. Furthermore, important data services, such as data protection, often require additional disparate software and/or infrastructure.
The single architecture approach — such as the Adaptive Flash platform from Nimble —eliminates storage silos because it is capable of meeting varying SLAs across a broad range of critical applications, on-the-fly.
BSM: Nimble Storage’s Adaptive Flash platform includes the ability to deliver all flash, auto flash, and disk-only service levels. For which types of applications are each of these best used?
Roth:
- All Flash: The entire dataset is served from flash, which guarantees deterministic low latency. This mode is ideal for latency-sensitive workloads. Examples are production online transmission processing (OLTP), Real-time analytics, virtual desktop infrastructure (VDI), etc.
- Auto Flash: This mode offers a balance of very high performance at a much lower cost of capacity. This suits applications that require both high performance and high-capacity. Examples include Exchange, SQL, SharePoint, Virtual environments, etc.
- No Flash: This mode is optimized around delivering low-cost capacity as data is efficiently stored on disk, using flash only for system meta-data. This is recommended for capacity-optimized workloads without high performance demands. Examples include backup, data repositories, archival, etc.
BSM: What is necessary to discuss with a client to determine the best type of storage for each application?
Roth: In Nimble’s case, this is extremely simple because we have a single platform (Adaptive Flash) and so our partners focus mostly on a) the amount of storage required and b) the overall performance requirement.
A good place to start is to get a general understanding of the performance and capacity requirements for each critical workload, along with a sense of future needs, if that can be estimated upfront. Some questions that should be discussed with the client include:
- How many IOPS does the application require?
- What are the latency requirements for the application?
- How much capacity is currently being consumed? How has this changed over the course of the last 3/6/9/12 months? How much capacity will be required 3/6/9/12 months from now?
- What type of data protection strategy (snapshot schedule based on recovery point objectives (RPOs) and recovery time objectives (RTOs)) is required by the application?
BSM: With the right combination of flash and non-flash, can clients save money while optimizing performance?
Roth: In short, yes. This depends on how efficiently storage resources such as flash are used, along with how the solution scales. A solution that scales performance and capacity independently — at the lowest incremental cost — allows organizations to forgo forklift upgrades and balance performance, capacity, and cost. If flash is being used efficiently, it is consumed only by active data; snapshots, RAID and parity data should be stored on disk, which is significantly cheaper. In addition, Nimble’s Storage-on-Demand program allows users to pay for storage on a utility basis, shifting from a capital expense model to an operating expense model.
Read the full press release with more information on Nimble Storage’s Adaptive Flash platform here.
To learn more about innovations within the Nimble Storage Adaptive Flash platform that allow enterprises to experience the benefits of storage consolidation without compromise, visit: www.nimblestorage.com/consolidation.