Anterovium - Fotolia
Small World Big Data
Published: 19 Oct 2017
Doubts about a cloud-based environment being little more than a passing fancy are vanishing. Plenty of real enterprises are not only comfortable releasing key workloads to public clouds, but are finding that hybrid operations at scale offer significant economic, productivity and competitive advantages over traditional on-premises data centers.
In fact, many of the big announcements at VMworld 2017 highlighted how mainstream businesses are now building and consuming hybrid and multicloud IT.
NSX all around
VMware has accelerated its transition from hypervisor vendor to cloud-management tool provider. Its virtual networking product, NSX, is not only a big source of revenue for VMware, but it also underpins many newer offerings, such as AppDefense, VMware Cloud on AWS and Network Insight. Basically, NSX has become the glue, the ether that fills VMware's multicloud management business.
By shifting the center of its universe from hypervisor to the network between and underneath everything, VMware can now provide command and control over infrastructure and applications running in data centers, clouds, mobile devices and even out to the brave new internet of things (IoT) edge.
More MaaS, please
VMware rolled out seven management as a service (MaaS) offerings. MaaS describes a sales model in which a vendor delivers systems management functionality as a remote, subscription utility service. MaaS is ideal for systems management tasks across multiple clouds and complex hybrid infrastructures.
One of the motivations for MaaS is that the IT administrator doesn't need to install or maintain on-premises IT management tools. Another is that the MaaS vendor gains an opportunity to mine big data aggregated across their entire customer pool, which should enable it to build deeply intelligent services.
Four of these new services are based on existing vRealize Operations technologies that VMware has repackaged for SaaS-style delivery. We've also heard that there are more MaaS products on the way.
It's important for vendors to offer MaaS services -- such as call home and remote monitoring -- as the inevitable future consumption model for all systems management. There isn't a single organization that benefits from employing an expert to maintain its internal, complex systems management tool. And with mobile, distributed and hybrid operations, most existing on-premises management products fall short of covering the whole enterprise IT architecture. I have no doubt the future is MaaS, a model that is bound to quickly attract IT shops that want to focus less on maintaining management tools and more on efficiently operating hybrid, multicloud architectures.
The VMworld show floor has been a real storage showcase in recent years, with vendors fighting for more attention and setting up bigger, flashier booths. But it seemed this year that the mainline storage vendors pulled back a bit. This could be because software-defined storage products such as VMware vSAN are growing so fast or that the not-so-subtle presence of Dell EMC storage has discouraged others from pushing as hard at this show. Or it could be that in this virtual hypervisor market, hyper-convergence (and open convergence too) is where it's at these days.
Maybe it's that all the past storage hoopla stemmed from flash storage crashing its way through the market. Competition on the flash angle is smoothing out now that everyone has flash-focused storage products. This year, nonvolatile memory express, or NVMe, was on everyone's roadmap, but there was very little NVMe out there ready to roll. I'd look to next year as the big year for NVMe vendor positioning. Who will get it first? Who will be fastest? Who will be most cost-efficient? While there is some argument that NVMe isn't going to disrupt the storage market as flash did, I expect similar first-to-market vendor competitions.
Data protection, on the other hand, seems to be gaining. Cohesity and other relatively new vendors have lots to offer organizations with a large virtual and cloud-based environment. While secondary storage hasn't always seemed sexy, scalable and performant secondary storage can make all the difference in how well the whole enterprise IT effort works. Newer scale-out designs can keep masses of secondary data online and easily available for recall or archive, restore, analytics and testing. Every day, we hear of new machine learning efforts to use bigger and deeper data histories.
These storage directions -- hyper-convergence, faster media and scale-out secondary storage -- all support a more distributed and hybrid approach to data center architectures.
Next-generation cloud management
In addition to birthing new MaaS options, VMware's cloud management tools, including the newly acquired Wavefront, expand with NSX-derived data. Independent system management vendors such as Embotics are also finding new running room and renewed opportunities, as enterprises finally embrace production multicloud strategies in earnest. Look for increased innovation in this area, including specific applications of machine learning, the use of ever bigger data sets, cross-cloud perspectives and hybrid operations with single points of management.
Everyone is talking about how to effectively manage cloud-native applications in the enterprise -- in other words, containers and container management. This market is young, but IT teams will need to manage container environments like they do VM environments, except with increased expectations for highly agile DevOps-style approaches. In this area, VMware announced Pivotal Container Service -- aka PKS, where the K stands for Kubernetes -- but the real story is yet to come on how IT will be able to effectively manage hundreds of thousands of temporal and temperamental containers.
Composing a cloud
Finally, I'm seeing real promise in dynamic bare-metal provisioning. Liqid has an interesting composable infrastructure based on its smart stack of software-driving Peripheral Component Interconnect Express switches. These switches can dynamically software-define physical servers, carving them out of racked pools of CPU boards, GPUs, memory cards and disks. If the target OS supports it, these switches can even dynamically add and subtract physical resources from a running OS. If you think about it, this seems the exact opposite approach to hypervisor virtualization when it comes to sharing a pool of physical resources.
Composable infrastructure is real. Hewlett Packard Enterprise's Synergy is already providing racks of dense composable resources to hundreds of businesses, although that offering is made up of proprietary hardware. There will no doubt be much discussion and exploration of use cases, and validation of relative costs and actual benefits. We can imagine that, very soon, both private and public clouds will support the on-demand and elastic cloud provisioning of physical servers and storage, not just VMs.
Future of the cloud
If a cloud-based environment and hybrid cloud management are finally becoming just part of normal IT operations, then what's the next big thing? Somehow, we'll need to dynamically migrate applications, users and data around these hybrid architectures to optimize spending while still assuring performance, availability and data protection. There will be a lot of dimensions to optimize.
Also, we will need to deploy, version and update data center blueprints as easily as application blueprints. Maybe that's the essence of the software-defined data center; not just infrastructure as code, but the whole stack of applications, services, security, service levels and ensuing infrastructure as code.
We'll also need to think not only about managing our hybrid infrastructure, but also IoT and industrial IoT value chains out to the edges of our enterprise world. If it has an IP address, we'll have to figure out how to support it.
Navigate different hybrid cloud management tools
Get tips on managing a hybrid cloud deployment
Avoid common hybrid cloud deployment mistakes