Microsoft hasn’t forgotten Windows Server. It’s about to get a big upgrade.
It’s been a while since Microsoft released Windows Server 2019, and development on the next major release is continuing. It’s still some way away, but if you’re opted into the six-monthly release schedule as part of a Software Assurance subscription it gets new features much more quickly. At Ignite 2019 Microsoft started to reveal what would come in the next Windows Server release, across several different conference sessions. The result is a picture of an OS that’s evolving to work with newer technologies, taking advantage of the cloud and focusing on Microsoft’s hybrid cloud vision, bridging the data center and Azure.
New features in Windows Admin Center
One of the key elements of the modern Windows Server environment is Windows Admin Center (WAC). Designed to support administrators working from desktop PCs, it’s a browser-based management environment that’s intended to replace Windows’ existing Remote Server Administration Tools and the built-in Server Manager, bringing Microsoft and third-party admin tools into one pane of glass that can target servers and desktops, both physical and virtual, on-premises and in the cloud.
One key development demonstrated at the Ignite event is a new performance monitoring tool that builds on the familiar Windows Server monitoring experience. You can now filter the counter list, to quickly build custom performance views, with access to real-time information from your servers. You can quickly put together groups of counters with contextual searching to help find relevant counters for the problem you’re trying to solve, along with analysis tools to help you explore your data and find the information you need. Once you’ve built an analysis workspace you can save it for future use and share it with colleagues and across servers, building a library of performance monitoring views.
Closely related is support for Azure’s Arc server management layer, which is intended to manage and deploy policies across your servers using Azure Resource Manager templates rather than Group Policies, securely running PowerShell on your servers to ensure policies are correctly applied.
Hyper-V is key to much of what Microsoft is doing today, on the desktop, on your servers, and in the cloud. Using Hyper-V on Azure helps prove on-premises scenarios, like live migration, using it to support hot-patching of servers. Moving running workloads between servers is now supported in Windows Admin Center, with the aim of WAC becoming a complete replacement of the existing Hyper-V Manager. It already has features that aren’t in Hyper-V Manager, for example the ability to group VMs and manage them together.
Microsoft is continuing to bring Azure features to on-premises Hyper-V. In a presentation at Ignite, Microsoft teased significant increases in the size of supported VMs, matching Azure’s 12TB of memory. That’s not surprising, as Microsoft has already announced support for 16PB of physical memory in the next major release of Windows Server.
SEE: Windows 10: A cheat sheet (TechRepublic)
Security remains important, and with CPU hyperthreading bugs affecting most of the modern generation of server CPUs, it’s important to be able to manage secure workloads so that they can be kept apart from untrusted workloads. Hyper-V is adding CPU groups to fix specific workloads to specific CPUs, allowing you to corral trusted workloads away from the rest of the applications and VMs you’re running. There are going to be improvements to how Microsoft is delivering its secure shielded VMs, although no details on implementation are available yet.
Using GPU-P in Hyper-V
GPU-based compute is increasingly important, and Microsoft has been adding support for it in Azure. The partitioning tools give multiple VMs access to a physical GPU, sharing it between the VMs. Each VM has full access to its GPU partition, giving it a significant performance boost without affecting the other VMs. If a VM doesn’t need GPU access, the partitions can be adjusted, ensuring that the GPU gets the best possible utilisation.
Moving applications to Windows Containers
Windows Containers are an important part of the future of Windows Server, acting as a new deployment target for your applications. Microsoft has spent the last couple of years rethinking the role of what was Windows Server Nano, using it as the basis for a stripped-down application host that provides the basis for Windows Container applications. It boots quickly, and provides the minimum services needed for an application. More complicated applications can take advantage of Windows Server Core, which has been substantially slimmed down, making it easier to customise for specific tasks.
Running Windows Server applications in containers makes them easier to operate in a hybrid mode, on Azure Stack hardware and in the cloud. With Windows Server a key element in Microsoft’s edge compute strategy, it’s an important role for Windows Server as both host and container OS. Hyper-V is a key part of the Windows container strategy, offering a thin, secure hypervisor to add further isolation between host OS and containerised applications.
SEE: 10 tricks and tweaks for customizing Windows 10 (free PDF) (TechRepublic)
Containerising workloads makes a lot of sense, especially where you have significant seasonal variability. The ability to burst compute capability across your data center and into the cloud is important, and it needs Windows Server Containers to support common applications without significant reworking.
Building out networking with SDNs
Building out virtual infrastructure on Windows Server has often required a significant number of VMs dedicated to virtual network appliances. Plans for future Windows Server releases aim to reduce the overhead needed to build and run a software-defined network (SDN), including making BGP (Border Gateway Protocol) optional. There’s a plan to make it easier to integrate Windows SDN tooling with Kubernetes, making it easier to build and run your own distributed systems, using Kubernetes tooling to manage the networking. Other networking features improve diagnostics, with the ability to model packets in a network. Microsoft plans to use WAC to bring SDNs and traditional networking closer together.
Using Azure Stack HCI Stretch for disaster recovery
With Azure Stack HCI (Hyper-Converged Infrastructure), Microsoft is aiming to address modern data center and edge compute scenarios, mixing familiar Windows Server 2019 features with elements of the Azure control plane. Azure Stack HCI can quickly build clustered servers, using off-the-shelf hardware from familiar vendors.
In the next long-term support release of Windows Server, Azure Stack HCI can be run in a ‘stretched’ mode across multiple sites for disaster recovery. Unlike the older failover cluster technologies, this approach is intended to be easy to setup and hard to misconfigure, using Windows Server’s Storage Replica features to link the sites. Nodes are automatically detected using Active Directory and IP addressing features in the OS, creating fault domains for the main site. The sites are configured with separate storage pools, using Storage Spaces Direct, with virtual disks at each site.
Microsoft is upgrading the Windows Server Health Service so it’s now able to work across distributed systems. If a server cluster fails, work will automatically transfer across to the second site. It’s a similar approach to that used when using Azure as a disaster recovery service, but keeps servers and data in your own data centers. The same underlying Storage Spaces technologies will be able to help on-premises data centers make better use of the available storage, as well as integrating with Azure storage services.
Getting ready for future hardware with PCIe 4.0
Hardware evolves and server operating systems need to evolve with it. PCIe 4.0 is the latest release of the PCIe standard, and it adds support for much faster data transfers, more than doubling speeds. It’s a technology that should have a significant effect on data center performance, providing driver support for faster networking and higher speed storage.