And now for the final pieces and a symphony…
Microsoft announced the RTM of Hyper-V™ on June 26th ’08. They had promised it would be ready within six months of the RTM of Server 2008, which would have been August ’08. At TechEd in early June, Microsoft announced that they were ahead of schedule for Hyper-V. They’d already gone into production, within Microsoft, with Hyper-V, for several months (e.g., all of the TechNet and MSDN web sites) with no significant (near zero) problems. So the quality is there – with the next question being, what about the performance? They wanted to get close to VMWARE with v1 of a major new architecture. The initial testing has exceeded their expectations – they were at or above VMWARE performance, even before release – credit the Microsoft Hyper-V VMBus and its “Synthetic” devices.
Hyper-V is the piece that will trigger customers moving to Server 2008. Hyper-V requires, but comes free with Server 2008. Installation is painless – it doesn’t require any complex training or complex steps. You start with a simple, normal, quick Server 2008 install – 2008 installs faster than ever, with all the distracting questions deferred to the post install – ICT, Initial Configuration Tasks.
After the normal server install, you just log on and add the Hyper-V “Role” using Server Manager – just like you would add “File Server” or “DNS” as a Role. The server then reboots and like magic, “jacks up the O/S” and inserts the “hypervisor” under the O/S – poof! done!…
The initial server installation is moved into a Hyper-V Parent partition. The Parent partition (Host) makes it easy to add/manage the Guest “virtual machines” (VMs) that will run in Child partitions. Hyper-V virtual machine (guests) management can be done using the Server Manager on the Parent – using the same Server Manager that was used to add the Hyper-V Role.
Once you gain experience with Hyper-V and Server 2008, you’ll want to take advantage of the option to install just the “Server Core” in the Parent partition, instead of a full Server install. Server Core being the “windows without windows” Server – perfect for production environments where you want and just need the “core” server – reduced resource requirements, fewer patches, fewer reboots.
We just need a “core” Parent/Host (mainly for its drivers). We can move back to our desktop for the remaining and full day-to-day VM management. Hyper-V doesn’t use the older (less than impressive) web interface console that we had with Virtual Server.
One option to manage the VMs is to use the new RSAT (Remote Server Admin Tools) – next generation “Adminpak” – and its Hyper-V mmc snap-in. Or step up to, and take advantage of, the current and future SC VMM (System Center Virtual Machine Manager) – the new SC VMM 2008 was announced in April and is currently in Beta.
…”System Center Virtual Machine Manager 2008 was designed to fully utilize the foundational features and services of Windows Server 2008 and Microsoft Hyper-V™ Server. This includes Hyper-V’s 64-bit architecture, attack hardened security model, and fail-over cluster support.”…
Since SC VMM has the ability to easily “snap-shot” a running, production system and convert it into a VM – P2V (physical to virtual) and then deploy the VM to your new Hyper-V host, it won’t be long before you have a bunch of production servers running as “guests” VMs on one physical host, better known as server consolidation. I’m sure that you bought a really reliable “brand name” server that would provide “availability through reliability”. But what if the hardware does fail or you’d like to step up to the world of HA – High Availability, the world of 99.99% availability, the four 9’s, the less than one hour downtime per year availability.
Server 2008 features a complete redesign of the 2003 Server Cluster option – renamed Failover Clusters. It’s still the HA solution to reduce downtime. But instead of the complex steps of a creating a cluster; then adding file shares or print shares, you can just add the “feature” clustering to an existing file server when you want to increase the uptime.
Even more significant (exciting) – Failover Clusters are fully aware of Virtual Machines, with both Host and Guest clustering options. A VM can be moved (quick migration) to a different Host or all the VMs can failover to a standby host, in the event of a Host failure.
“…In VMM 2008, creating a high availability virtual machine (HA VM) has never been easier. Gone are the complex multi-step manual processes from before – now, an administrator clicks a simple checkbox which designates a VM as highly available. Behinds the scenes, VMM orchestrates the creation of that HA VA which includes instructing the Intelligent Placement feature of VMM 2008 to recommend only hosts that are part of a host cluster for the newly minted HA VM. “….
So now we have a symphony of:
Failover Clusters, and