Data Center Modernization has definitely reached critical mass. The message that came from TechEd 2013 was “It’s time to make Hybrid Cloud Real.” That, of course, starts with the modernizing your data center to be able to implement private clouds. On top of that, more and more data centers are migrating their hypervisors to Hyper-V in spite of the greater footprint a full Windows Server operating system has on the bare metal. The feature parity as well as cost savings that comes from Hyper-V as a feature (and the subsequent removal of the VMWare tax) offsets the hassle of the additional footprint.
Windows Server bare metal hosts running Hyper-V, like other hypervisors, support SAN boot of the operating system drive using iSCSI. It is important to realize that the iSCSI services depend on the underlying storage and iSCSI network being provisioned properly to accommodate the eccentricities of how Windows boots from SAN using network interface cards in place of traditional storage adapters or HBAs.
Understand the Supportability Parameters
The supportability of the storage support comes from the storage vendor. This also extends to iSCSI boot SAN scenarios per the KB article: https://support.microsoft.com/en-us/kb/305547/en-us. Even though the article does not mention Windows Server 2012 (or R2) it is still in place. Normally, this would not be complicated but in the case of iSCSI networks, the device may likely be using a NIC to locate the storage (especially if they are actually using NAS – network attached storage – i.e. NetApp) and not a traditional storage adapter or HBA.
Slipstream your 3rd-party drivers if possible
The use of slipstreamed NIC/Storage drivers in the installation ISO will prevent any timing issues from swapping back and forth between driver media and OS media. The may be especially the case if you are controlling headless blade devices using KVM or some other solution. I have found that this resolves many of the issues outlined in this particular KB: https://support.microsoft.com/en-us/kb/2826787 – as well as the 0x80070057 error message when trying to format drives or create partitions during the operating system setup.
No Thin-Provisioning LUNs for the OS Boot Drive
LUNS on the NAS devices (i.e. NetApp Devices) need to be thick provisioned for the drive containing the OS instead of thin-provisioned. In addition LUNS for the host OS boot volume only should be 127GB or less. Remember this is only in the context of the LUN being used for host devices iSCSI boot volume.
Avoid using Default Gateways for iSCSI NICs
The NICs configured for the iSCSI SAN should avoid having a default gateway. This can cause issues such as slow throughput occurring during formatting of disks and the copying of files during installation. This has been an issue with the Windows iSCSI initiator in the past and has previous appeared in KB articles:
960104: If you start a system from iSCSI, the gateway specified in the iSCSI Boot solution will always be used by Windows to communicate with the iSCSI Target
2727330: Default gateway is set to 0.0.0.0 if you start a Windows Vista-based, Windows 7-based, Windows Server 2008-based or Windows Server 2008 R2-based computer from an iSCSI boot device
In addition, the network ports connecting to the boot volume iSCSI interfaces on the iSCSI network’s switch should have ICMP redirect disabled.
If all else fails . . . revert to the old way!
If the interactive installation still fails, remember – there is the legacy way of deploying Windows Servers in an iSCSI SAN boot configuration outlined in:
Over the past 20 years, I have had many different roles in IT. I’ve been a helpdesk jockey, professor/instructor, sysadmin, developer, support engineer, escalation engineer, and now consultant. I’ve worked with a variety of industries as well. I’ve been both a customer and an employee of a Fortune 100 software company. As I have moved into various roles through my career, I’ve simultaneously watched the growth of the IT community in pontificating in various mediums ranging from community forums to full-blown tabloid tech journalism. I’ve learned what kind of statements garner respect and attention and what are often dismissed as hyperbole or sensationalism.
The bloggers are supposed to represent the users and/or IT pros – the “pulse” of the community. In many cases the quality of the bloggers are positive as they derived excellent content and insights due to one or more of the following factors:
Experience: A blogger will likely be taken seriously if they have the experience to back up what they are talking about. This is why the best insights often come buried deep inside of community forums and not necessarily on the site of a full-time blogger or tech journal. Why? Because blogging is not their job. They ARE an IT Pro. Blogging is merely a hobby.
Depth of Analytical Thought: They demonstrate an outstanding aptitude for critical thinking. Even if the source is focused towards a specific vendor (or as many say – biased) the analysis is spot on.
Depth of Technical Thought: Simply – they know the technology inside and out. They yield a wealth of technical information and for that reason alone, they often command respect.
I am here to tell you the influence bloggers have on software vendors and products often depend on how they engage and embrace the community around the vendor and its products – regardless of how they may “bash” a product or feature or “praise” it. If the community respects the blogger, their stature increases with the software vendor. If the blogger is simply ranting or spilling out hyperbole for the sole purpose of “click-bait,” that can come back to haunt them. This is often a challenge for full-time bloggers who are often selling advertisements to generate revenue or perhaps are freelancing for a journal who pays them literally by the click.
When you build up that large amount of overhead you need to keep those clicks and ad views going, the blogger has no choice but to be a provocateur to remain relevant in the IT tabloid media that those same bloggers helped to create. When an IT analyst or an IT research firm publishes opinions or assessments, they are always taken more seriously as they represent a wealth of combined experiences and knowledge bases. They approach product, technology, and industry analysis in a much more scientific and data-driven process. The research firms publish both the analytical and the technical depth in every case.
Since most major software vendors, at least in the US, are publicly traded, it is Wall Street that ultimately has the most influence on its direction. in IT, your shareholders are often your customers as well.
I’d been wanting to write an article on this subject for a while, but this week, I was inspired to the write this article after reading three distinct articles relating to RDS/VDI –a technology I worked in extensively. I have the unique opportunity to cite examples of an attempt of influence by a blogger, a group of analysts, and a group of investors in a very busy week for the VDI industry.
http://www.brianmadden.com/blogs/brianmadden/archive/2015/06/10/server-os-based-vdi-is-an-official-quot-feature-quot-of-windows-server-2016-apparently-microsoft-plans-to-continue-screwing-us-for-years-to-come.aspx – Basically, Brian Madden still hates how Microsoft does VDI. In other news, the Sun came out this morning.
The Analysts: http://www.gartner.com/document/3072722 – a brutally honest assessment by Gartner on why VDI is not ready for the cloud and what it will take to get VDI to a true cloud-based DaaS (Desktop as a Service.)
The Investors: http://www.valuewalk.com/2015/06/citrix-systems-inc-ctxs-elliott-associates-letter/ The investment Group Elliott Management reveal their desires for change at Citrix (the leader in VDI) in an open letter to its CEO and Board of Directors.
Which of those three articles that I mentioned do I pay the most attention to? Well, I always trust analysis over hyperbole – but money trumps all.