A very revealing case study just published by VMware describes how the cumbersome management of Hyper-V prompted a former customer to return to VMware.  Dierbergs — a supermarket chain in the Midwest — began their virtualization journey in 2011 with VMware vSphere, but later succumbed to the temptation of seemingly lower cost Hyper-V and invested considerable effort in migrating over to the Microsoft hypervisor.

In the study, IT Infrastructure Manager Chris Lindloff highlights key frustrations they had with Windows Hyper-V:

Dierbergs experienced five major outages in a single year, taking down critical machines necessary for inventory and product ordering. “Even with the highest level of Microsoft support, we had difficulty resolving every outage,” says Lindloff. “Some of our outages lasted as long as 12 hours. One outage even corrupted a critical database server that we needed to restore from backup, resulting in significant data loss.

No surprise here — since Hyper-V is tightly dependent on Windows, Patch Tuesday applies not only to the applications running in the data center but also to the entire virtual infrastructure – double trouble!  VMware vSphere takes an entirely new and different approach by using a very lightweight, purpose-built hypervisor as the foundation for a resilient software defined data center.

Technology leadership at Dierbergs made the critical decision to move back to VMware technology this year:

…while Hyper-V promised to be less expensive, that simply wasn’t true in practice. The total cost of ownership was far higher because of the need for additional management time and extensive support.

If your IT team is spending excessive cycles troubleshooting random connectivity issues or poring over disparate recommendations in an attempt to avoid patching disasters, consider the real-world experience of Dierbergs CIO Jim Shipley:

With the VMware solutions, we spend more of our day focusing on strategic initiatives, which in turn helps IT drive real value back into the business.

The hypervisor is a critical component of the data center, not a commodity as challengers would have you believe.  Standardize on VMware vSphere so your technology pros can focus on the future, and the days of experimenting with alternatives will soon be in the past.

Tags: ,

Don’t fall into the trap of believing that everyone else has managed to architect a successful multi-hypervisor private cloud — the trend is actually heading in the opposite direction.

A recent survey by Wikibon found that multi-hypervisor environments are actually projected to decline to 43% next year, down from 52% today.  Wikibon also concluded:

… the experimentation phase of multi-hypervisor is coming to an end, and consolidation of the number of hypervisors is expected over the next few years. When asked about their strategy, 48% of the sample said they were willing to trade the risk of lock-in for simplification and expect to focus on a single hypervisor approach.

For the most part, those advocating a deliberate multi-hypervisor strategy do not necessarily have the end user’s best interest in mind.  They are either looking for a way to increase their own market share or sell a solution to help manage the complexity of a heterogeneous environment.

Microsoft, which has been trying desperately for over half a decade to break out of the mid-20% virtualization market share range, would like nothing more than to get a foot in the door of established VMware shops.  Which is why they invest heavily in attempts to lure VMware professionals over to Hyper-V:

The “over 70%” figure referenced above likely comes from an ESG survey that does not attempt to explain the proportional share of the hypervisors in question.  There is no denying that complex data centers are rarely homogenous — despite the burden of added management overhead, there are often multiple vendors of server hardware, networking equipment, operating systems, and application frameworks.  One would not expect the hypervisor to necessarily be an exception.  However, consider the reality that even in an environment with hundreds of VMware ESXi hosts, it takes just a single Hyper-V host to trigger the “multiple hypervisors” designation.

Lastly, consider this pithy, if not slightly facetious, perspective from Gartner analyst Chris Wolf on the topic:

cswolf-tweet-doomed-to-fail.png

The hypervisor is a crucial element of the virtualized data center and hybrid cloud. Reliability, manageability, and security benefits ensure VMware will continue to dominate as customers everywhere count on VMware vSphere for their primary virtualization platform by an incredible margin.

Tags: , , ,

The HP blade server ecosystem is very flexible, especially when it comes to I/O connectivity modules, and one popular choice for Ethernet is the Cisco 3020 Blade Switch.  While the HP C7000 does help reduce the nest of physical cables found in a typical server rack, vSphere administrators are on their own when it comes to managing the individual blade switches — nothing new here, just another bespoke Cisco-powered network handcrafted by skilled artisans. It’s not uncommon to have four separate blade switch modules in a chassis, each corresponding to one Ethernet interface on an HP blade server.

Wouldn’t it be nice to have an easy way to quickly view the status of of all four interfaces connected to an individual blade server without having to telnet into four different blade switches and run IOS commands?  Enter clogin, a command-line utility that comes with the Really Awesome New Cisco config Differ (RANCID) tool suite, and should be packaged for most Linux distributions.

If you’re using Mac OS X, good news: RANCID is available through Homebrew.

Once the package is installed, create a .cloginrc file (with 600 permissions) in your home directory with at least the following line to enable seamless authentication to the switches:

Now test a simple command to see if things are working:

if that is successful, you can create a simple shell script to automate logging into all four blade switches to get stats for a particular blade server.  For example:

Run the script and pass the blade number (typically from 1-16 on a C7000) for a quick glance at the status of all ports as well as the VLAN info:

It’s not exactly software defined networking, but sure comes in handy when troubleshooting connectivity issues with blades.

Tags: , , ,

One of the great things about Mac OS X is the ability to use a traditional shell — Bash — as the native command line interface.  However, those accustomed to the rich environment of a Linux Bash shell may miss many of the handy CLI tools that don’t typically come with Mac OS X. Thankfully, there is a very easy-to-use package manager called Homebrew to fill this niche.

Dependency Alert

Homebrew makes use of the Apple Command Line Tools, which are handy additions to your Mac as well.  Downloading the tools is easy but requires registering as an Apple Developer first — it’s fast and free.  After logging in, look for something like this:

Download Apple Command Line Tools

Homebrew Installation in a Snap

After installing the Apple tools, installing Homebrew is as easy as this one-liner:

Here are some handy management commands to get started:

Once Homebrew is set up, it’s easy to add useful utilities to your Mac; to browse an interactive catalog take a look at Braumeister.  Here are some suggestions to try:

  • nmap
  • mtr
  • dos2unix
  • watch
  • wget

An interesting characteristic of Homebrew is that in most cases it actually downloads source code and compiles it for your system.  I see this as generally positive, using the genuine version of an app instead of a one-off binary ported and built by an individual.  Having said that, a feature known as bottles enables binary packages to be distributed.

Another appealing design attribute is that Homebrew is quite safe to use — it installs everything under /usr/local/ and symlinks from appropriate locations:

Now that you’re enhancing the Mac command line, it’s time to really up the ante…

Bash Completion on Mac OS X

The Bash shell has a very versatile and productivity-boosting completion mechanism.  By default it works like most command shells, automatically expanding directory and file names by hitting ‘tab’ – but there is much more potential, such as completion of hostnames when using ssh.  Once Homebrew is up and running, simply do this to enable completion for a broad range of CLI tools:

Homebrew bash-completion

Edit your .bash_profile as described above to enable completion so you can just hit ‘tab’ to let the productivity begin!

Tags: ,

There are a few different ways to install VMware Tools on Linux.  The quickest is to use the Operating System Specific Packages (OSPs) — an online repository of packages built for certain popular Linux distributions.  When deploying Ubuntu in the datacenter, it makes sense to choose the Long Term Support (LTS) releases — version 12.04 “Precise Pangolin” as of this writing, which has OSP support.

It is easy enough to install the OSPs manually, but it’s even better to automate the process along with the OS deployment.  Ubuntu uses Debian “preseeding” for automated installs, which is similar to Kickstart.

The following preseed directives will add the VMware Tools OSP repository, import crypto keys, and install the “nox” tools, which are appropriate for GUI-less servers (be sure to remove or modify the http_proxy setting for your environment):

Perform an automated PXE install and your vSphere VM will be ready for service:

Ubuntu-OSP-tools

Tags: , , ,

« Older entries § Newer entries »