Randomness (via

New Cisco Catalyst 4500-X 10GE modular switch

Along with 40 GE and 100 GE modules, Cisco recently announced the Catalyst 4500-X 10GE aggregation switch.

Cisco Catalyst 4500-X family

10GE Density

On its face, it looks similar to the Cisco Nexus 5000 series switches, but looking at the features it is definitely more of a campus LAN switch, rather than the Nexus data center line. There are two base configurations - a 16-port and a 32-port 10GE models. Each model has an expansion slot that currently supports an 8-port 10GE uplink model (Cisco datasheets suggest that a 40GE uplink module is on the roadmap).

Similar to the Nexus 5000s, the 4500-X ports support SFP+ 10GE optics along with 1GE SFP modules.

Layer 3 Support

The 4500-X supports both IPv4 and IPv6 routing in hardware, along with support for VRF-Lite and “Easy Virtual Network” (EVN) features. (The Nexus 5000s require an additional expansion module for layer 3 support).


The most intriguing feature in the 4500-X may be built-in VSS (“Virtual Switch System”) support. Two 4500-X switches can be linked by 10GE ports and configured as a single logical switch. This simplifies configuration while providing a higher level of availability. It also allows ether-channels to be be built across two switches (for link redundancy and performance while eliminating the need to build spanning-tree triangles). The VSS feature has previously been reserved for the 6500 chassis with Supervisor 720s. The 4500-X therefore offers a much more cost effective way to provide a highly available distribution layer (or even a core for smaller environments that want a 10GE backbone).

4500x as a distribution.


The 4500-X is an intriguing new solution as an aggregation switch for campus LANs that want to bring in 10GE uplinks without the cost or complexity of a chassis based switch. VSS capabilities in particularly allow for a dual switch redundant solution that logically functions as a single switch — a solution that up until now has required a much more expensive chassis based switch.


Dealing with Disaster

Michael Lopp has a nice post about how he has learned to manage “sky is falling” situations. If you work in IT, you have been there (and as a long time consultant I’ve been there at least twenty times too many), but have you handled the situation effectively?

The tendency all too often in these situations is to “shotgun” it, try anything and everything, because everyone is breathing down your neck and it needs to be fixed RIGHT NOW. In many cases though this can make things worse.

As Lopp says:

Action feels like progress, but undirected action is not progress, nor is it a plan. You’re going to barge into the office and start barking orders because that is what everyone expects, but if your orders are not shaped by what you’re really attempting to do, you are just scurrying people around aimlessly. Yes, you get lucky. Yes, everyone breathes a sigh of relief when you show up with your impressive sense of purpose, but in my experience when my direction doesn’t map to intent, I’m usually getting no closer to propping up the sky.

At the root of the situation is the fact that you very likely have an incomplete picture of what the real problem is. The person you are directly interacting with is probably riding you about their specific issue and you get hyperfocused on a symptom or secondary effect, while not clearly seeing the big picture. There are likely other symptoms that aren’t being communicated effectively.

The trick then is to resist the urge to act on each and every impulse and instead take the time to get a complete picture of what’s really going on.

This takes incredible DISCIPLINE, precisely because so many people are anxious to have a resolution RIGHT NOW.

So, as Lopp suggests, take the time to establish a “war room”, a central base for collecting information. Your “war room” may be as simple as a single whiteboard, or a sheet of paper. Create a diagram of the systems impacted. On the side, add a list of reported issues (likely symptoms) and a distribution list [1] for sending updates. If the issue is big enough of course you will likely set up a true “war room” as a central point of data collection, and where various troubleshooting activities can be managed from. Be patient in collecting that information until you have a good mental model of the situation.

A sorry attempt at flowcharting the war room
A sorry attempt at flowcharting the “war room”

Have patience to act only when your plan has been vetted by people you trust and a course of action becomes clear. Yes, this will be hard. Yes, you will have customers, co-workers, and bosses anxious to do something. But, act too quickly with incomplete information and you likely will end up making the situation worse.

In the face of disaster, it’s the wise person who does not act until they know. Unfucking the situation is a bandaid, understanding what you’re truly trying to fix is a cure.

- Klaus

  1. Divide the distribution list into “tactical” contacts – people who are actively involved in troubleshooting, testing, or otherwise need regular technical updates; and “executive” contacts – those who only need summary updates on overall status. Include anyone who shows interest or expresses “need to know” to the appropriate list.  ↩


Top Virtualization Blogs

Check out’s 2012 top virtualization blogs poll results. Lots of good resources there. In addition to the overall list, blogs are also listed by categories if you are only interested in say storage, or VDI.


Microsoft Office? Good Riddance!

There was speculation the other day about leaked information that an iPad version of Microsoft Office is in the works and due to be released soon (along with an intriguing thought that Microsoft might share the stage with Apple at its upcoming iPad event to announce it).

My thought? Pshaw... I hope it never comes to the iPad. It really is about time for Office to lose its stranglehold on our work.

[Begin Rant] I really do find Microsoft Word to be the most anti-productive "productivity" software I've ever used. Especially for an ADD pseudo-perfectionist like myself. I can barely write two word without stressing about styles and formatting. And then I try to paste something in from a website or another doc and my styling goes all haywire. Drives. Me. Bonkers! Always has. [End Rant]

Unfortunately in the corporate world today it is darn near impossible to avoid Word. But I'm attempting to minimize my Word usage until it's absolutely necessary. I've been inspired quite a bit in this endeavor by people like Merlin Mann and David Sparks (want to go down the geek rabbit hole for several hours? - check out Merlin's two Mac Power Users episodes) . The answer for now, though far from perfect, for me is plain text and Markdown (and some custom hacked together CSS and cool apps like Marked).

The reason I like this approach is that I can have all my styles mapped out in a CSS file, and when I need to write I can just pick a text editor, any text editor, and write. To be honest I end up using MultiMarkdown more than pure Markdown since it also lets me easily make tables, and add footnotes, and other cool stuff. When I'm done, I can use Marked to export the styled text into a PDF or HTML and have something that is all nice and formatted that required minimal fiddling while actually writing it.

This "text only" approach is made even easier these days by the proliferation of Markdown and Dropbox enabled text editors for iPhone, iPad, and Mac (much harder to find on Windows however - I'm biased here, but I really do feel like the best software these days is on the Mac) - apps like PlainText, Nebulous Notes, nvAlt, Byword, and Sublime Text.

There is one area where this all falls apart pretty quickly, and unfortunately I still need a good solution for, and that's getting all this writing back into Microsoft Word (it's just not possible to avoid in corporate America). There are ways to get Markdown formatted text exported to HTML and RTF, but then I still end up having to fiddle with styles and formats to get it all to look right again, but at least I can save the fiddling until I'm done with 99% of the content.

Of course when there's a problem and a geek all it takes is a weekend or so and we'll come up with a solution... (now if I only had a free weekend).

- Klaus

PS. This post was written using Markdown in Byword.


VMware PEX 2012 Field Notes

It’s been a fun and educational week at VMware’s Partner Exchange (PEX). PEX is like VMworld but limited to VARs, System Integrators, and other partner organization - with a mixture of technical as well as sales focused sessions from VMware and other vendors, including Cisco, HP, EMC, and Netapp, discussing their virtualization related solutions .

VMware saves its big new product announcements for VMworld so there is not much to news to share, but there was a lot of great information on existing solutions including great detail on design and implementation best practices. My goal was to get a more well rounded understanding of vSphere 5, vCloud Director, Site Recovery Manager, and View. Here are some of the things I learned.

vSphere 5

The deployments I have been working with are still primarily based on vSphere 4 and 4.1. Version 5 isn’t radically different[1], but it does introduce a few components that suggest the way forward for vSphere - in particular VCSA and Web Client.


VCSA, or vCenter Server Appliance, is a new Linux [2] based vCenter Server appliance. It is not yet at feature-parity with the traditional Windows version, but for smaller greenfield deployments its a nice alternative. It can be installed with a built-in DB2 or an external Oracle database. The biggest benefit right now is ease of installation - it is much easier to install the VCSA OVF than to have to install and configure a Windows Server, then install SQL server and vCenter server. Among its limitations are lack of support for linked moded, SRM, and ESX/ESXi versions prior to 4.0. I wouldn't be surprised if future releases will come closer and closer to the capabilities of the Windows-based vCenter.

Web Client

VSphere 5 also has a much improved web-based client. As with VCSA there is not yet feature parity with the Windows desktop client though it is clear that VMware is also heading in that direction. The web client as it currently stands offers a nice interface for occasional admins that have been delegated some responsibilities but don’t need the full power of the desktop client.

Site Recovery Manager (SRM) 5

The most talked about feature in SRM 5 was vSphere Replication (VR). This feature allows customers to replicate individual VMs between sites independent of the underlying storage arrays. In fact, since the replication is handled entirely by the vSphere software, it can utilize two totally different storage arrays (e.g. The primary site may have an EMC array, and the secondary site may have a Netapp SAN).

The biggest downsides to VR are that replication is only asynchronous with a 15 minute RPO [3] at best, and there are currently no capabilities for fail-back. The SAN independence however makes this a great option for lower-tier applications, even if array-based replication is used for mission critical apps.

VMware View

I don’t know that there was much new about View, but there was certainly lots of good information about best practices for View deployments. This will certainly come in handy as I expect we will continue to see strong demand for this in the coming year. A big theme of the executive keynotes was “Millenial Entitlement”, or better stated, how new expectations of interactions with applications and data are beginning to drive IT department’s behavior. View and other VMware solutions will begin to focus on facilitating a shift in this direction as well.

vCloud Director

With vCloud Director and ancillary products such as Chargeback, VMware is certainly hoping to make it easier for IT departments and organizations to deliver “IT as a Service” (IaaS) to its customers (be they external or internal). There are a lot of details that are beyond the scope of this (already very) long–winded post, but it certainly opens a realm of new possibilities for IT departments in all organizations.


There was also a big focus at PEX on manageability, particularly with VMware Operations Manager. I personally didn’t get much exposure, but several collegues did. It’s obvious that this will also be a big focus for VMware in the coming year.

The Way Forward

VMware CTO Steve Herrod suggested three areas VMware considers strategic for future investments and innovation:

  • integrating into the “post-pc” era (see also “Millenial Entitlement”)
  • vSphere as a platform for application development
  • cloud management - making self service and service delivery easier - think of each IT department being able to offer its own “App Store”

More to come

I have many more notes that I expect I’ll feel motivated to expand on in the near future. Thanks for sticking with me this far!

- Klaus

  1. There are certainly lots of new features including significant increases on parameters such as supported memory and datastore size that will allow vSphere to scale more gracefully.  ↩

  2. SUSE Linux based  ↩

  3. ”Recovery Point Objective” - a recovered server may have 15 minute old data  ↩