I recently read a Redmond Magazine article in which writer Kurt Mackie discussed the top trends for IT professionals to pay attention to over the next five years, as presented by Gartner Inc.’s David J. Cappuccio. I am not one to drink the Gartner Kool-Aid, but this is an interesting list, and there is a lot of content that is relevant to the IT pro that is working in a Microsoft-based infrastructure. So here, I present my thoughts on Gartner’s list of the top 10 IT trends, as given from the Microsoft POV.
Anything that is hardware-defined is inflexible. As anyone who has worked in the hosting business will tell you, flexibility is your friend, and cloud computing yearns for the ability to have agile computing. Traditional networking lacks in numerous ways.
By using networking that is software-defined, you move to a deployment that can be automated, easily modified, and scale beyond the limits of hardware. SDN is available in both Windows Azure (Virtual Networks) and in Hyper-V since Windows Server 2012 (Hyper-V Network Virtualization). HNV is now in its second release in Windows Server 2012 R2, and when combined with System Center 2012, it is a real-world technology that you can deploy right now. With HNV you provision fewer VLANs, purely for the cloud infrastructure, and then create VM Networks for your tenants that contain Virtual Subnets. Virtual machines and services running within the Virtual Subnets can route within a VM Network. They can also route to and from the physical network via an NVGRE Gateway, which can be either a third-party appliance or enabled within Windows Server 2012 R2 virtual machines.
Once again, software-defined storage (SDS) is a solution for the cloud, be it public or private. There are several benefits to SDS, including:
There are two aspects to Windows Server 2012 R2 that can be viewed as SDS. The first is the Scale-Out File Server (SOFS), which allows you to build a SAN-like architecture from modest clustered Windows file servers and some form of failover cluster-supported storage, such as Storage Spaces, which offers very economic disk capacity while offering great performance potential. Storage Spaces aggregates and virtualizes physical disks and the capacity is shared via the SMB 3.0 protocol. If you know Failover Clustering, file shares, and Windows networking, then you’re pretty close to knowing how to build a SOFS.
Shared VHDX allows virtual machines to be clustered without the virtual machines owner needing visibility of the cloud infrastructure. Being a VHDX file, this virtualized form of cluster-supported storage is suitable for self-service.
Microsoft is in a very unique position when it comes to hybrid cloud. This is because Microsoft is the only company to offer a solution for private cloud (Windows Server and System Center), public cloud by the partner (Windows Server, System Center, and Windows Azure Pack), and public cloud by Microsoft (Azure, which is based on Hyper-V). Microsoft’s Cloud OS in the unification of those platforms into a hybrid cloud, allowing a customer to deploy a service on any or all of the three clouds. That allows the customer to choose the most suitable service for any tier of a service.
Microsoft is making a huge push on Azure, and one of the messages that they want to have customers understand is that the hybrid solution is the best solution, linking on-premise, partner-managed hosting, and Microsoft’s public cloud.
You want integration? Microsoft has integration coming out of the woodwork. The components of System Center are deeply integrated natively and via Orchestrator. System Center completes Windows Server. Active Directory integrates into Microsoft’s public cloud services, such as Windows Intune and Office 365, through Windows Azure Active Directory. Windows Intune integrates mobile and remote/bring-your-own (BYO) device management with System Center Configuration Manager’s (aka ConfigMgr) enterprise devices. Integration in Microsoft solutions is topic that you could write about for weeks!
With this topic, Gartner is referring to having a more rapid but managed release cadence for applications, especially across mobile devices.
ConfigMgr was the first solution from Microsoft to offer managed self-service application delivery. Administrators can manage an internal “app store” that users of enterprise PCs can browse and select applications from. Free or site-licensed applications can be installed immediately without waiting on a helpdesk response. Those applications requiring budget spend can be requested and a workflow can be initiated (with the greater integration of System Center) to allow the employee’s manager or department budget owner to approve the installation.
Windows Intune has a similar system, where Microsoft’s cloud based mobile or BYO device management solution can download apps from an official public app store or from a company app store to Windows, Windows RT, Windows Phone, iOS, or Android devices.
Microsoft has believed in this pervasive computing concept in which devices of all kinds are connected to the Internet. This could be your car, a pacemaker, a watch, or just about anything, really! ZDNet’s Mary Jo Foley recently reported that Microsoft has been staffing up a team on The Internet of Things.
The Open Compute Project aims to develop more energy efficient data centers. Thanks to cloud computing, data centers are popping up overnight like mushrooms. They suck in huge amounts of energy, and we all know that energy costs are only ever going to increase.
Microsoft’s Global Foundation Services is responsible for building some of the most ingenious energy efficient data centers on the planet. If you ever get the chance to take a tour, do so! You won’t be allowed to talk about what you saw, but you’ll be impressed. In a drive to be more efficient, and therefore more price competitive, Microsoft has developed practices and designs for a leaner cloud. Microsoft has joined and started to contribute their learnings to the Open Compute Project.
Gartner also calls this concept a software-defined data center. Clouds grow rapidly and they cannot wait for humans to deploy infrastructure, diagnose and repair faults. Workloads must also be mobile across data centers in the event of a regional disaster.
At TechEd 2012 North America, Microsoft started to flesh out their concept of a Datacenter Abstraction Layer (DAL). This is a concept where you build up a compute platform template and deploy it to a rack, or racks, of bare metal hardware. System Center 2012 R2 can reach out to servers via baseboard management controllers (BMCs) to build Hyper-V hosts (compute) and SOFSs (software-defined storage). Top-of-rack (or access) switches can be configured, servers configured with networking, and SDN in the form of Hyper-V Network Virtualization configured for virtual workloads. In other words, you can deploy a rack or pod full of servers and storage, and System Center can turn that into new cloud capacity.
Windows Azure has the concept of regional data centers for fault tolerance. For example, data stored in the North Europe data center in Ireland replicates asynchronously to the West Europe data center in The Netherlands (wait – the West facility is actually east of the North one?!). Hyper-V Replica also builds in asynchronous replication into on premise clouds based on Windows Server 2012 and later.
“Do more with less” is a message that most IT pros are familiar with, and this is set to continue for the foreseeable future. There will be pressure for more efficient hardware, in terms of compute power, the cost of acquisition, the cost of ownership, capacity, and licensing.
If you’re legally licensing (and cost effectively) Windows Server workloads as virtual machines then you are already licensed to run Hyper-V. More and more stories are appearing of large enterprises and hosting companies making the jump to the Microsoft stack. Hosting companies are running proof of concept projects on SOFS storage because of the much lower cost:TB ratio.
A public cloud, such as Windows Azure, also gives software developers and testers (common in large enterprises) a way to rapidly deploy and destroy virtual machines without making a long-term commitment to capital investment for additional private cloud infrastructure. This allows IT to focus budget on production systems that run on premise, reduces spending pressure, and makes the testers happier. By the way, that all can integrate thanks to the concepts of the Cloud OS hybridization and a common compute platform.
Change leads to fear. Fear leads to anger. Personally, I don’t see IT objections to change as something that will happen in the future – they’re already here and they have been all the time. Cloud computing scares IT pros, their IT managers, and consulting companies because it outsources work to hosting companies. In my own experience, infrastructure-as-a-service (IaaS) mostly impacts hardware resellers; the need for software development, installation, configuration, and support will continue. Remember how Office 365 initially faced objections? That software-as-a-service (SaaS) product now has gained acceptance and has disrupted (in a positive way) the traditional deployment of Exchange, SharePoint, and Lync on premise. In fact, we are now starting to see that the SaaS products are creeping ahead of their on-premise alternatives!
It seems to me that Microsoft has a pretty good grasp of the future, or at least, the future that Gartner sees.