DRP Guide Part Two: Elements of an Effective DRP
Data Center Maintenance
This is leading more vendors toward offering specialized products for block, file, and object storage.
The good news for the channel is that storage is hot. In the InformationWeek 2018 State of Infrastructure Study, storage beat out even cloud services as a top factor driving IT infrastructure change. Additionally, 62% said their data volumes are increasing by more than 10% annually, while few (7%) reported hosting the bulk of their storage (75% or more) in the cloud.
The takeaway is that enterprises and SMEs have a lot more data to host, and they are going to need new hardware to do it.
The real news isn’t the capacity, however, it’s the shift in expectations. Traditionally, storage was primarily about archiving, compliance, and disaster recovery. With the rise of big data analytics and machine learning, data has taken on strategic value, which businesses of all sizes are intent to squeeze for all its insightful juice. This trend is pushing customers to prioritize storage performance to a degree not common in the past.
Thus we find, for example, “unified storage” is losing some luster. The Jack of All Trades approach doesn’t permit the storage software necessary to excel in each use case. This is leading more vendors toward offering specialized products for block, file, and object storage.
Among the types of products most likely to meet customers’ needs today and tomorrow are:
- All-Flash arrays
No one who has installed SSDs is regretting the decision, and there is good reason to believe materials shortages are abating. Costs will start to fall again, leaving few barriers to installing Flash for performance and OPEX savings reasons. Impressive transfer rates of 10 Gb per second will become the standard. An associated outcome, interest in hybrid arrays will decline, as there are few compelling arguments for complicating systems with the HDDs.
- Storage packed with NVMe SSDs
These low-latency interfaces add about 10% to the purchase price of a storage system but deliver two to three times the performance and about half the latency, so customers are getting ready to dive in. That being said, NVMe is also likely to show up in most product lineups, so it won’t be a significant brand differentiator.
- Hyperconverged infrastructure (HCI)
IDC predicts the HCI market will grow to almost $5 billion next year; that’s up from $2 billion in 2016. The software-centric approach and almost Lego-like scaling is a good fit for many data centers. HCI with All Flash is an obvious choice.
- Storage class memory (SCM)
Support for SCM was included in Microsoft Windows Server 2016 and most Linux kernals, so the world is ready. But price point compared to Flash (SCM lines up better against Flash than DRAM) may keep a lid on demand for now. However, 10x read/write speed over Flash, higher IOPS, comparable throughput, and granular data access at the bit or word level will tempt some enterprises in this direction. SCM may be more of a 2019-and-beyond concern, but the channel should begin preparing now.
- Products with an in-cloud equivalent
HPE Cloud Volumes and IBM Spectrum Virtualize are just two examples. Arrays are more often coming packaged with cloud equivalents for replication or failover.
Most products appealing to the enterprise, at least, will need to enable software-defined storage, which will increasingly be delivered in containers to provide for flexible programmable infrastructures. And rapidly emerging, machine learning will facilitate new storage management processes to optimize workload services.
Manufacturer-wise, Dell-EMC, HPE, IBM, and NetApp still have a place, but All Flash has opened doors for newer entrants, such as Pure Storage, and HCI is propelling sales for the likes of Nutanix and others upwards.
The Channel Challenge
Storage is becoming more performance-driven, and the ecosystem of OEMs is becoming more diverse. This picture presents a number of challenges to channel partners in terms of delivering, implementing, and supporting the storage technologies customers are now and will soon be clamoring for.
For one thing, much of the storage growth is shifting to new market entrants excelling in specific areas. Think Nutanix for HCI. There are simply more brands and more products to be cognizant of. This makes it harder for a channel partner, especially one tied strongly to one or two major OEMs, to offer and support a sufficiently enticing variety of technologies to keep customers at the trough, rather than shopping elsewhere for those proficient with various emerging best-of-breed manufacturers.
The shift in some areas from unified to specialized storage for each layer adds to the complexity and will require additional engineering depth on the implementation and support bench. Heading in the opposite direction, HCI means mixing compute and storage in new ways, which can necessitate a generalist’s knowledge to address certain troubleshooting tasks. Add in software-defined storage and AI-based workload optimization and channel partners will need to leverage both hardware and software understanding to deliver the performance and support customers expect. The right mix of engineering staff will be critical, but all the more difficult to retain while the talent gap continues to gape.
As storage shifts to strategic applications, winning channel partners will need to demonstrate the ability to drive performance gains in storage installations. From a support perspective, swapping out a Flash drive and other straightforward break/fix calls are now the equivalent of a commodity server. It’s hard to differentiate there.
The stand-out channel partners and other support providers will have the capabilities to hone fine points of performance and solve problems arising in more complex data center architectures. Those that fail the test will find customers looking elsewhere.
This positions premier third party maintenance partners well for the future of storage support. The breadth of talent a large, support-focused vendor can marshal will tend to overshadow that available among channel partners. With many OEMs trying to limit the use of their own Level 3 resources, encouraging channel partners to handle initial support calls, customers with performance concerns will see value in companies that deliver immediate Level 3 engineering assistance without the wait times and escalation procedures. Additionally, predictive maintenance and proactive fault detection takes on greater import as performance takes center stage. Combining these capabilities with quick-turnaround spares fulfillment for the more standard break/fix calls creates a full support package.
That this recipe for support success is well known doesn’t make it easy to replicate, especially at the 60% cost-savings Gartner found for third party maintenance over OEM and channel support. Although customers will no doubt continue to purchase support contracts through the channel, a switch to third party maintenance after an initial period is attractive for the simplicity across the data center, the price point, and the responsive, high-level engineering capabilities.