A TRANSFORMATION IS UNDERWAY in data centers around the globe as organizations transition from legacy storage solutions to all-flash storage arrays. However, companies must be mindful that putting more storage capacity in a smaller space increases the need for high-resiliency in their storage solutions.
To be sure, all-flash provides significant benefits in the data center, starting with a lower cost per input/output per second (IOPS) as compared to traditional disk-based storage systems. Flash also offers the opportunity for storage consolidation because of its greater capacity, scalability, and predictably high performance.
Companies also get greater efficiency with flash storage systems, with the ability to store more data in a smaller footprint, creating opportunities for consolidation of storage systems because of real-time data reduction. And all-flash systems are easier to manage, with little need for load balancing and tuning to remove I/O hot spots.
But the move to all-flash data centers does not come without some challenges and an element of increased risk of data loss. The idea that flash storage can handle more data in a smaller space means companies are putting more volume, hosts, hypervisors, and applications on a single piece of infrastructure. It’s similar to the move to virtualization years ago, when companies began running increasing numbers of applications on a single server outfitted with multiple virtual machines (VMs).
While such consolidation is great from a resource utilization perspective, it increases the risk of data loss should the data center suffer a disruption, whether from a natural disaster, power outage, human error, malware, or the like.
As organizations move to all-flash data centers, they need to adopt advanced recovery tools that can provide data protection without sacrificing the superior performance they seek from their flash investment. In short, flash arrays require organizations to put a greater focus on lowering data loss risk and increasing array resiliency.
Drivers for all-flash
Businesses increasingly view their data as a corporate asset and are constantly trying to extract more value from it. For many companies, effective use of data can even be a competitive differentiator. Recent surveys of CEOs and IT decision makers by Frost & Sullivan find that 65% of businesses believe managing data growth is important to business success. And nearly half (45%) say competitors’ use of data and analytics holds the greatest potential to disrupt the industry.
Against that backdrop, businesses are looking for their storage solutions to deliver on a range of requirements, Frost & Sullivan finds.2 First, they must be able to deal with multiple file types and formats, including fast-growing unstructured data. Storage solutions must allow access to the same data from multiple applications, including latency-sensitive and high-performance apps that require near real time performance.
Data must always be available, which means a data protection solution must support more aggressive recovery point and recovery time objectives (RPOs/RTOs) with all-flash. Storage systems must also recognize that data is increasingly subject to continuous analytics, and support the various business intelligence platforms companies are adopting.
All-flash data center requirements
All-flash data centers going forward need to factor in the following requirements:
The first necessity is application acceleration or superior I/O performance. Originally, performance was the main reason customers were interested in flash; now it’s table stakes. Whether for databases, containerized applications, or traditional or emerging apps, a flash storage system should deliver faster and more predictable response times than traditional storage solutions.
Operational simplicity is another requirement. The flash storage solution must be managed along with the rest of the data center infrastructure, from whatever management tool the organization already uses, fitting in seamlessly without adding an extra burden.
The flash solution should also offer investment protection. At a base level, that means it should support technologies such as in-line data compression and deduplication, enabling companies to maximize utilization of flash media. But it should also be able to work with advanced technologies such as non-volatile memory express (NVMe), which reduces latency and increases IOPS. Emerging storage-class memory technology (also known as persistent memory) is another consideration, as it promises great leaps in speed and data resiliency. The point is: Customers must ensure their chosen flash storage solution comes with a roadmap that gives some comfort level that it will keep up with advances in storage technology over time.
Finally, the all-flash storage solution must provide a high degree of risk mitigation, with support for data replication and recovery systems that offer complete protection, and can meet the RPOs and RTOs that the business requires to ensure near-continuous data availability.
Providing effective data protection
In searching for a data protection solution that can meet these various challenges, companies should consider several criteria.
First, to ensure good performance, the data protection solution should support storage snapshots for rapid backup. Traditional full or incremental server backups typically only happen once per day and can be a drag on performance. In high-availability environments, storage snapshots provide more frequent, point-in-time copies of data that enable organizations to meet tight RPOs and fast RTOs. Because they are dealing with relatively small amounts of data, snapshots and differential snapshots won’t affect application performance.
Hewlett Packard Enterprise (HPE) provides fast allflash storage arrays, and its partnership with Veeam Software enables the company to bring enterprise-level resiliency to customers with all-flash data centers.
For its part, HPE offers Tier 1 3PAR StoreServ All-Flash primary storage systems along with the HPE StoreOnce System for secondary storage, which delivers extremely high performance and classleading deduplicating efficiency. The two are tightly integrated to minimize complexity and reduce total cost of ownership.
HPE solutions support application integration, which accelerates performance and deduplication efficiency while minimizing RTOs for applications such as Microsoft Exchange, Active Directory, SharePoint, SAP HANA, Oracle, and SQL. On their own, HPE solutions are highly resilient and secure, but by integrating Veeam software, they deliver another level of data protection (see Fig. 1). Veeam provides availability for virtual, physical, and cloud-based workloads with several important capabilities.
HPE & Veeam in action
Consider the results London’s Gatwick Airport has seen with the HPE-Veeam combination. Mindful that any downtime would result in passenger delays and perhaps data loss, the airport seeks to maintain 24/7 availability across its two data centers, which house a mix of virtual and physical servers. If any VM fails in one data center, a mirror image in the other data center takes over. But Gatwick wanted to be able to recover the original as quickly as possible to restore redundancy and resiliency to its operation. Restoring VMs with traditional backup tools could take hours or even days.
Veeam certainly helps PGA Superstore IT staff meet the expectations of its business executives. As Anderson puts it, “With a solution like Veeam, we know we can tell them that ‘it just works.’”