The How and Why of High-Performance Primary Storage as-a-service

The key to reducing primary storage costs is to better manage the data that is on primary storage. The problem remains that IT lacks the tools, and more importantly the time, to properly manage the workloads that primary storage supports. Without the proper management, costs for primary storage are prohibitive while also increasing the cost of secondary and backup storage. Primary Storage as-a-service (PSaaS) promises to lighten IT’s load in terms of making storage accessible on-demand while ensuring performance is high. At the same time, PSaaS can reduce secondary and backup storage costs as well as potentially eliminate the need for additional backup and data management software solutions.

PSaaS, like any other “as-a-service” offering, is fully managed. A fully managed solution means that IT no longer has to spend time tuning performance, provisioning volumes, or even handling tasks like data protection and archiving. With PSaaS, IT professionals can focus on tasks that have a more meaningful impact on the organization. Now, IT professionals can focus on innovation, instead of managing infrastructure.

What’s NOT Primary Storage as-a-service?

IT planners need to be careful as they consider PSaaS. Many vendors claim to offer an operating expenses (OPEX) purchasing model for primary storage, but PSaaS is more than a buying methodology. In most cases, so-called PSaaS offerings are traditional capital expenditures (CAPEX) based, on-site storage solutions using leasing as a disguise. The organization is still responsible for all day-to-day storage management and data protection tasks. They also need to purchase separate solutions for executing these tasks. Capacity expansion typically requires a minimum time-in-use commitment, and organizations can’t quickly scale down their capacity requirements for a reduced service rate. There is also no reduction in on-premises infrastructure requirements.

What IS Primary Storage as-a-service?

True PSaaS is available with very minimal upfront commitments, and organizations can scale down their capacity requirements without penalty. While there is some IT involvement, PSaaS also automates many of the day-to-day storage management and data protection tasks. No additional solutions are required to perform these tasks. They have the capability to perform them automatically. The goal of PSaaS is to provide as much on-premises storage performance as the organization needs, harnessing the edge and the public cloud and minimizing the data center footprint.

How Does Primary Storage as-a-service Work?

The architecture of PSaaS is critical for the offering to deliver on its seemingly conflicting commitments of both high performance and a small footprint. At the heart of the solution is software that overlays the hardware and unifies separate storage locations into a single capacity pool that IT can consume at will. The software needs to provide both block and file access so all of the organization’s workloads can leverage the service. Finally, the software needs to provide complete management and protection of the data it stores, across all layers.

The software then works across its network that combines the on-premises cache, a point of presence at the edge and the public cloud to maximize performance, minimize the data center footprint and lower costs. At each of the organization’s data centers, the service provides an all-flash, on-premises cache that provides low latency, high-performance storage IO for applications. The on-premises cache stores the most current data set and to minimize data center footprint, as data ages, it seamlessly moves data across location from on-premises, to the regional POP and to the cloud. Most organizations access less than 10% of their data on a day-to-day basis, so the on-premises capacity and footprint savings can be significant.

It is critical though, that all locations can respond quickly to data not stored on the on-premises cache. Transparent response across locations enables the organization to embrace the concept of only keeping the most active data on-premises and accessing the rest via high-performance connections to the cloud. Any perceived performance loss as a result of data not behaving as if it were on premises, may lead the organization to abandon the concept.

The second location, often missing from PSaaS, is a warm storage layer, typically located at the edge, in a regional point of presence (PoP). Edge location is critical to creating a successful service and getting stakeholders to embrace the PSaaS concept. While the service optimizes the on-premise cache for maximum effectiveness, there are times when users or applications request data not in the cache. Most multi-location solutions typically try to service cache misses exclusively from public cloud storage. The public cloud generally is geographically too far away and, as a result, has too much latency to provide the response that a production application requires. The in-region, edge-based PoP can respond to an IO request in milliseconds, which is plenty of time for most applications and workloads.

The third location, the public cloud, serves as an archive tier as well as an offsite backup for all data, without requiring replication. Because dedicated networking throughout the service and the connectivity of the on-premises cache to the regionally-based PoP at the edge and then to the cloud, the third location can harness the benefits of the cloud and use the least expensive storage available. And, thanks to public cloud snapshot technology and data center replication, the third location meets almost all of an organization’s requirements for data protection, retention and access.

Is Primary Storage as-a-service the Right Prescription for Your Organization?

PSaaS, if configured as described above, is suitable for a wide variety of workloads. Core data center applications like VMware, Oracle, and MS-SQL, as well as modern applications that support today’s IoT and digital media demands are all viable candidates. An edge-based network and on-prem entry point make accessing the service immediate while minimizing costs while also supporting block, file and object workloads.

It’s important to understand the integration between the on-premises appliance, edge-based PoP and the public cloud. The traditional approach of buying as much flash capacity in the on-premises device as the organization can afford goes away in a PSaaS model. In the PaaS model only the most active data, and potentially near-active data, is stored in an on-premises flash-based cache. The colder data is stored in the cloud, but as we discuss about the right PSaaS technology also resolves latency by using an intermediary PoP at the edge. As a result the organization can trust it’s won’t suffer a performance impact even in event of a cache miss.

Getting Started with Primary Storage as-a-service

Since PSaaS is a subscription-based offering, the organization can start small, learn how the service operates and gradually expand and move workloads over to the service as it makes sense. Most organizations start with one workload that is either new or creating challenges on-premises or whose storage system is up for a refresh. The organization can get started easily with minimal migration work required. In fact, the on-premises entry point to the service presents itself as a standard block or file device, so it operates the same as the original system it is replacing.

Once the migration is complete, storage is now consumed on-demand and data can be accessed when and where it’s needed. The organization should run the first workload for a few weeks, long enough to for the data placement algorithms to learn access patterns. During this time, users are able to experience cache misses to see if it is impactful or even noticeable. The first weeks of use should be the worst-case scenario. During this time, the data placement algorithms are learning data access and the chances of a chance miss are at their highest. The PSaaS solution’s predictive algorithms become more accurate over time, and the workload’s performance should continue to improve. Next, IT can select new workloads to add, on-demand, adding storage via the web-based portal and requiring no extension of the existing on-prem footprint. In most cases the organization is able to move the majority of its workloads to the PSaaS with no or minimal expansion of the on-premises cache device.

Unexpected Benefits of Primary Storage as-a-service

One of the most significant impacts of PSaaS is how it changes the secondary storage and data protection infrastructures. The most frequently used data is stored in the customer’s data center, while hot and warm data are stored at the edge in a PoP and all data is backed up to the public cloud. All the data, regardless of location, is transparently available to users and applications.

Second, the solution provides complete and automatic data protection. Organizations may decide that the PSaaS’ protection can replace their current protection software and hardware. All data is protected by snapshots and a complete durable copy that is stored in the public cloud. IT can choose to keep the snapshots for varying times up to indefinitely. The snapshots are immutable, which protects the organization’s data from ransomware.

In a disaster, the organization has two options. In the first one, it can implement a secondary on-premises entry point in a new location and allow the cache to re-populate. Once deployed, their applications and workloads are instantly available. Performance may suffer while the on-premises device is re-hydrated, but applications are available. It is essentially a “streaming” recovery.

The second DR option is to implement a version of the appliance in the public cloud and leverage cloud compute to host its applications and workloads. By using PSaaS the organization is essentially getting disaster recovery as a service (DRaaS) included with the same solution. IT can also use this implementation variant to use the public cloud for non-DR use cases like dev/test or cloud bursting. Workloads or copies of workloads can run unchanged in the cloud. The cloud version of the on-premises appliance provides a native environment from which they operate.

Conclusion

Primary Storage as-a-service solves a lot of the challenges organizations face with their current on-premises storage infrastructure. However, IT planners need to be careful to make sure the service is genuinely a subscription, and not a typical on-premises solution disguising itself as a service. IT planners want to look for true cloud integration that limits the on-premises storage footprint and leverages the cloud to store backup data, older data, and protected copies. Also critical is the use of the edge and a local Point of Presence that ensures a warm cache layer to deliver the low latency that applications require. A solution that meets these requirements should improve overall performance, lower storage costs, and promote an organization’s data protection and disaster recovery capabilities.

Sponsored by ClearSky Data

Sign up for our Newsletter. Get updates on our latest articles and webinars, plus EXCLUSIVE subscriber only content.

Twelve years ago George Crump founded Storage Switzerland with one simple goal; to educate IT professionals about all aspects of data center storage. He is the primary contributor to Storage Switzerland and is a heavily sought after public speaker. With over 25 years of experience designing storage solutions for data centers across the US, he has seen the birth of such technologies as RAID, NAS and SAN, Virtualization, Cloud and Enterprise Flash. Prior to founding Storage Switzerland he was CTO at one of the nation's largest storage integrators where he was in charge of technology testing, integration and product selection.

Tagged with: , , , , , , , , ,
Posted in Article

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 22,236 other followers

Blog Stats
  • 1,554,254 views
%d bloggers like this: