Enterprise IT is an increasingly cloud-aware undertaking—more and more applications and data are migrating to Amazon.com AWS, Google Cloud Platform, and Microsoft ’s Azure. It’s important to note though that not all workloads are following suit with the widespread cloud migration. This fact is driving the industry to focus on supporting hybrid cloud solutions—solutions that allow workloads to be split between cloud-only and on-prem compute and storage in an organization’s datacenter. Hybrid cloud solutions exist to make migrating between an organization’s datacenter and a public cloud provider painless. Every OEM today seems to have some flavor of a hybrid-cloud offering.
An increasing number of cloud services that exist to bridge the public cloud suppliers and proprietary, on-site, solutions are evolving to supplement hybrid-cloud offerings. Nowhere is this more interesting than in the storage-as-a-service space. The performance and locality demands of storage (relative to compute) make it very difficult for non-cloud providers to pull off effectively.
All of the public cloud providers have block and object storage as part of their offerings, each with more than acceptable performance and latency for most workloads. This makes adding value beyond the native cloud providers a tricky proposition. The value proposition that third-party storage providers tout revolves around their ability to provide a cloud-provider agnostic offering. Third-party providers can support a customer’s data simultaneously across a range of cloud vendors, all while providing traditional storage array features into those environments. IT organizations get cross-cloud capabilities, with a storage management model that they understand and have experience using.
Hewlett Packard Enterprise Cloud Volumes
There are a number of vendors playing in the storage-as-a-service space. Zadara Storage
provides block, object and file storage as a service, both on-premises and across cloud vendors. Wasabi Technologies
has an interesting offering for S3-style object storage-as-a-service. In the tier-one storage OEM world, NetApp
has partnered with Microsoft
to provide its NAS functionality to Azure clients. NetApp
also offers a preconfigured virtual appliance for AWS customers, but does not have a standalone storage-as-a-service offering.
The offering that has me most intrigued is from Hewlett Packard Enterprise
(HPE). HPE isn’t known for its cloud or software-as-a-service offerings, which makes the HPE Cloud Volumes a bit of an anomaly in its line-up. HPE does offer its GreenLake pay-as-you-use services, but those are implemented using on-premises equipment and aren’t hosted by HPE directly. HPE Cloud Volumes, on the other hand, is a purely hosted service.
The hosted approach makes sense for storage, given that data should be as local as possible relative to the compute elements that are consuming that data. This keeps latency low, which is the overarching goal of nearly every storage solution. HPE says it often hosts its storage arrays in the same co-location facilities as the cloud providers themselves.
When the compute instances are running in the same regions that HPE is serving data from, HPE Cloud Volumes offers throughput and latencies that are largely indistinguishable (and, in some cases, better) than the cloud providers’ own solutions. HPE today hosts within the major east and west coast regions, as defined by both Microsoft and Amazon.com AWS. Hewlett Packard Enterprise says it is following customer demand as it ramps up its own service and will move to other regions as that demand expands.
HPE Cloud Volumes came to HPE as part of the Nimble Storage acquisition in mid-2017, and HPE has chosen to keep the service alive. It’s based on the HPE NimbleOS, which powers HPE’s Nimble-branded storage arrays. This allows HPE Cloud Volumes to leverage a mature storage stack while providing block storage services to applications running on both Amazon.com AWS and Microsoft Azure (it should be noted that HPE doesn’t yet support the Google Cloud Platform, but says it will follow customers if demand increases).
An organization’s cloud-based compute resources can access the Hewlett Packard Enterprise Cloud Volumes using the same native interfaces that it would to access native cloud volumes (e.g. an HPE Cloud Volume looks like an Amazon.com AWS EBS volume to an AWS compute instance). This is implemented using ExpressRoute on Amazon, and VNET on Microsoft Azure. The ability to switch a volume between Microsoft and Amazon hosted clouds provides a great deal of flexibility for a customer’s end-solution.
HPE Cloud Volumes also allows the ability to use the HPE Cloud Volume as a replication target for any HPE Nimble Storage systems that you may have in your existing IT infrastructure. This is powerful. It allows an easy migration from existing storage into the HPE Cloud Volume and isn’t something that is achievable with any of the native cloud volume offerings available today. This make makes it extremely compelling for existing HPE storage customers.
HPE Cloud Volumes is a very good first offering in the storage-as-a-service space. It stands out in feature set and performance. As Hewlett Packard Enterprise continues to evolve it, it has the potential to be something special. While there is a rich feature set today, there are a few things that would make it an even more compelling and competitive offering for enterprise and SMB customers alike. Snapshots using S3 as a target (something we would also like to see in HPE’s standalone arrays), an object store interface, and deeper integration with HPE’s stellar InfoSight predictive analytics would all go a long way towards making HPE Cloud Volumes an unbeatable product.
An IT decision maker’s dilemma
Leveraging storage-as-a-service for cloud-hosted applications is a complicated choice for an IT organization, especially when the cloud providers offer very acceptable native solutions. It becomes an easier choice when there is a desire to share storage resources across cloud vendors, or even between a single cloud provider and on-premise servers. It’s an easier decision still if the service takes on the feature set found in traditional storage arrays—features such as replication and snapshotting, storage virtualization, and the like.
HPE has a compelling solution on its hands. HPE Cloud Volumes is built around existing enterprise-grade, mature storage products, backed by a technology organization with a broad and deep portfolio of storage technology, and supported by an unmatched support organization. This big question is this: how serious is HPE about storage-as-a-service, or is HPE Cloud Volumes a one-off experiment, inherited and maintained as part of an acquisition? The answer to that question is what customers need to be comfortable with before deploying a business-critical solution on HPE Cloud Volumes. HPE talks and seems committed, but at the same time this is a product that has been in its portfolio for less than a year and sits apart from most of its existing portfolio.
We’re anxious to see Hewlett Packard Enterprise evolve this product. It really does have amazing potential—potential that would give HPE the opportunity to stand alone among its peers.
Note: This blog contains contributions by Moor Insights & Strategy principal analyst and founder, Patrick Moorhead.