Windows Server 2012 R2 – what’s new?

NOTE: while I’m still keeping the current posts live as they still seem to help, currently my focus has changed and new activity moved to the new site iternia.be

Microsoft just announced the R2 update of Windows Server 2012. Although 2012 was already another big step from 2008R2 and very feature complete, there’s always room for improvement… and the previous 3 releases I must admit I have been very pleasantly surprised each time with more ease of use, features and stability.

So what can we look forward to? These are my favorites:

1. Further stability and performance progress with HyperV

  • Some annoying shortcomings compared to market leader Vmware are finally getting crushed: previously memory thin provisioning was already a major jump forward but now we’re also getting live virtual disk expansion and shared ISOs should no longer block live migration.
  • Shared VHDX is paving the way for virtual clusterdisks. Finally we can backup clusters through the hypervisor with products like Veeam as the current setup with iSCSI disks still had to be backed up through pain-in-the-ass agent-based backups or storage snapshotting while the rest of all your servers were nicely snapshotted and backed–up with a 100% success ratio.
  • It is a public secret that HyperV is currently only ready for use by the SMB-segment, not high-end enterprise, but this is slowly changing. Both HyperV server and the management tools (SCVMM) are showing a lot of progress since the latest stream of bug-crushing hostfixes a few weeks ago. We’re running HyperV server 2012 + SCVMM 2012 SP1 (since the beta release in november 2012 and the RTM early this year) and finally the much needed fixes are available to get it acceptably stable.
  • Overall performance improvements are certainly welcome to improve the guest-to-host ratio and the steady increase with every update and fix shows HyperV is going in the good direction.

2. HyperV replica + HyperV Recovery Manager

  • Using Veeam in our datacenters so can’t fully judge HV-replica but I’m quite sure this technology will serve a lot of users well. I’ll try to give it a testrun soon.

3. HyperV Live migration data compression

  • Compressing a live migration stream before pumping it into a sometimes too small pipeline (f.e. a costly WAN-link between 2 datacenters) is certainly welcome! Hopefully the increased load on the host doing the compression won’t impact performance for the other guests. We’ll see once we get a testversion… For us, offsite replication from our main datacenter to our disaster-recovery site is an expensive operation. If we can save on bandwith that would be great and I’m sure many of you will think the same.

4. HyperV network monitoring

  • Wireshark for VM-guests 😉 Sounds good because troubleshooting some problems with wireshark requires installation inside the OS, including replacing network drivers. Not really an operation that passes rigid change management easily…

But it’s not all guaranteed succes. I’m not so sure about the following:

1. Storage spaces: again a decentralization of storage? We got great enterprise solutions from EMC’, NetApp, IBM, Whiptail and co. Perhaps the large CAPEX might in some scenario’s be worse than the somewhat more OPEX focused pizza-boxes approach but still in the end I doubt that it will be more profitable for the same performance-to-money ratio. I also have some doubts as to whether this will not create new problems, like…

  • SAN appliances are very specialised, feature complete and their OS strippped of unnecessary features for stability. Windows is still in the first place an alrounder and as we’re drawing storage back towards the same ecosystem as the applications => there’s an increased chance of storage failures due to faults in services that should not be able to impact the storage service. Applicative services must only rely on, but not directly relate to storage services unless it is only their own storage they can impact.
  • Microsoft is not a storage vendor. They already have difficulty to remain up-to-date with storage technologies; f.e. they support iSCSI and fiberchannel for mounting SAN-clusterdisks but the fiberchannel-over-ethernet variant is after all these years still not supported (although it actually works). Microsoft has a huge portfolio with OS’es, end-user-applications, server-applications, security-software (and recent years is becoming more and more a hardware vendor too). But whether this is good for the overall quality of each product is a difficult question. The integration and interoperability between products is certainly great but some products just lack the constant upward spiral needed to keep them top of the market.
  • Technologies aren’ always accepted as standard and remain a “Microsoft only” feature.

2. NVGRE network virtualization

  • The underlying technology is ready, but the standardization and support are not.
  • There is limited support from network hardware vendors and leveraging software, f.e. NVGRE-gateways.
  • NVGRE is not certain to survive. There are at least 2 competing technologies: VXLAN and OpenFlow. Each solution has a part of the vendor ecosystem behind it and new info suggests everyone might be talking about trying to improve and standardize OpenFlow although that’s far from certain either.

Hope we’ll be able to get a preview/beta version soon so I can get you an update 😉

To be continued…

Advertisements

About Geert Baeten
IT service architect - cloud infrastructure solutions - datacenter infrastructure solutions - service design / governing processes

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: