It’s always fun to make these kind of predictions. If only for the purpose of allowing me to look back in a year’s time and see how well, or poorly, I can predict the future. So, here goes…
#1 – OpenStack Overlays Get Noticed
OpenStack is now on it’s 14th release – Newton – and is on a very aggressive 6 month release cycle. More and more projects continue to be contributed back to the code base adding much needed functionality and ease of deployment/manageability. A few established companies have joined the fray by creating their own distributions of OpenStack – RedHat, Canonical, Mirantis, etc. – and have allowed many companies to adopt the platform by providing support services and training. While this has been beneficial for the Fortune 1000, medium and smaller size companies have been much slower to adopt. Even with support and training, OpenStack requires special talent to administer and maintain. This is getting addressed as the product matures, but it has left a market gap.
With this gap, a few entities have entered the market with management tools and abstractions that will allow customers to use OpenStack without really needing to get into the weeds of administering each of the OpenStack components. A few of these that have made great strides are Platform9, Cloudbase, and Stratoscale. I believe all of these will make great strides in 2017 and bring greater adoption of OpenStack to the marketplace.
Introduction to Platform9 with Sirish Raghuram from Stephen Foskett on Vimeo.
#2 – Continued Push for Hybrid Cloud Solutions
I think we’re beyond speaking in terms of absolutes when it comes to the use of cloud-base solutions. Most companies will choose a route that allows them to consume some hyperscaler-based resources, but still look to keep some workloads on premises. This brings with it some challenges. How do we stitch networks together to allow these applications to talk to one another? How do we ensure data security, data ownership, and data privacy based on where data resides? How do we provide for data/workload mobility from/to on-premises from/to hyperscaler? How do we protect data when the processes to do so are so incredibly different between data residing on a hyperscaler and data residing on premises? All of these questions will get answered in 2017 in some way, shape, or form.
Three major hybrid solutions really standout as gaining more attention in 2017:
- Azure Stack – Microsoft’s program to bring Azure to an on-premises solution (sounds weird, don’t it 🙂 ) will grow in maturity and gain in popularity for customers already using Azure as their hyperscaler of choice. When you pair something like Azure Stack with Azure ExpressRoute, the prospects get even more interesting. This could be an incredible solution to allow customers to seamlessly use static resources in a colo and burst with hyperscaler-based resources as needed. One of the biggest barriers-to-entry with Azure Stack has been the hardware platform. Even as of this writing, the hardware-compatibility requirements are very tight. If Microsoft loosened those up, this is a solution that could gain incredible adoption.
- VMware vSphere on AWS – The announcement of this solution back in October was explosive. VMware has wisely decided to scale back their vCloud Air solution and partner with a hyperscaler behemoth such as Amazon. Specifics on this solution are still vague, but it gives one the ability to deploy a vSphere cluster on AWS baremetal servers via a web portal. This opens up the possibility of customers spinning up vSphere clusters on-demand for dev/test, disaster recovery, or any number of other scenarios. When you pair something like this with solutions like NetApp’s Cloud ONTAP (a virtualized version of NetApp’s on-premises FAS arrays), it gives clients the option of replicating their data to AWS using array-based tools they already know and not have to pay for the DR virtualization/compute until it’s needed. (NOTE: Folks at NetApp have told me that they’ve already validated NFS and iSCSI connectivity from Cloud ONTAP will work with this solution). The official release will be in July of 2017 and we’re really looking forward to how this will play out.
- OpenStack Omni – At OpenStack Summit Barcelona in October 2016, Platform9 announced they were contributing their internal hybrid cloud project called Omni back to the OpenStack community to be a community driven project. Omni was originally developed to allow resources in a hyperscaler – specifically Amazon Web Services – to be orchestrated and managed through OpenStack Horizon and the other OpenStack component APIs. The project has only been public for a few months, but it does bring a lot of promise to stitching on premises OpenStack deployments with the hyperscalers.
#3 – Use of Object Storage Increases…
…driven mostly by backup and recovery software support. We’re seeing this more and more. Backup vendors like Veeam, Veritas, and Commvault are adopting the use of object storage APIs (mostly S3 or Swift compatible) to allow customers to push backup and recovery data to either an on premises object storage array and/or to cloud-based solutions. We’re even seeing this being built into on premise storage arrays as a way to tier cold data to cheaper storage. This becomes incredibly important when we look at customers hosting their data on All Flash disk arrays. NetApp has quickly adopted this approach with their use of FabricPools. This will initially allow customers using an All Flash FAS disk array to tier off snapshot data to something akin to StorageGRID or something similar.
#4 – SaaS Backup Solutions Will Become a Requirement
Remember the Salesforce.com outage from back in May? I sure do. And it has, almost single-handedly, created enough concern with customers that many are looking for ways to protect the data hosted on Software-as-a-Service solutions such as Salesforce. This includes email on solutions like Office365 and Google’s G Suite. It also includes project management information on solutions like Trello, Asana, or Basecamp. Even scarier is when you consider how all of the data within group chat solutions like Slack, Hipchat, etc. is being protected. These concerns will bring forth new and novel ways to protect data on each of these platforms. Not to shock anyone, but I would be remiss if I didn’t mention how NetApp is addressing this. NetApp announced back at Insight this year a new partnership with Amazon Web Services to provide a service directed at the largest of these SaaS offerings. The utility is called CloudControl and will be announced more formally in early January of 2017.
#5 – Hyper-Converged Architecture Will Become Less Clear
Not that it was all that clear to begin with, but HCI will become a murky architecture. As of today, people look at HCI in a fairly clear-cut manner – it’s storage, compute, and networking built into a single appliance and scaled out by clustering the appliances together. Today’s HCI gives the promise of easy management through a single (or two – Nutanix) management interfaces. It focuses on the rise of IT generalists within customer organizations – IT administrators that must have multiple skill sets and can ill-afford to be an expert in just one or two areas. That last comment is most important, not necessarily the architecture. There are multiple ways to provide ease-of-management to an IT generalist without necessarily needing to collapse an entire datacenter into an appliance-based architecture. One could argue that the OpenStack overlays discussed in point #1 could fit that need. To keep you thinking more along those lines, check out the following vBrownBag Opening Acts panel discussion from this year’s VMworld:
In Conclusion…
It ought to be a very interesting year. I concentrated heavily on data center related technology and didn’t mention much of the areas within BigData where I also focus. I see a number of new things coming on that front as well, although I’m less certain of specifics. Security will be front and center with certainty. I see projects such as Apache Knox, Apache Ranger, and Apache Atlas gaining more traction in the Apache stack. Offerings like Splunk or the ELK stack will need to get in front of this as well. I also see network telemetry analytics gaining in momentum as solutions like Cisco’s Tetration brings market visibility to the technology. I’m less bullish about Tetration in and of itself. I see solutions like Splunk or even emerging projects like Apache Spot as gaining greater adoption as they both have lower economic entry points (Spot being an opensource project).
With that, Happy New Year to everyone and I sincerely hope you have a wonderful 2017!