John Simmons' vBlog

I think it’s pretty common for engineer type minds to resist the new or unusual. Working in the IT bubble I find many colleagues have an inherent resistance to change. New or less-common technologies and techniques have a tendency to be written off. The ‘if it ain’t broke, don’t fix it’ attitude is not without value but is pervasive and limiting. What about some of those technologies that break the mold and fall ever so slightly outside of peoples realm of comfort? They can’t all be bad. In fact, some of them are extremely cool and have CRAZY cool uses.

Is this fair? I don’t think so. (Disclaimer: The Philosopher in me recognizes the problems with the concept of fairness)

Infiniband is a great example of one of these technologies. I recently heard an individual say, “I will never have Infiniband in my data center”. I don’t know the motivation for that specific prejudice, but it is not a unique stance. One of the coolest technologies out there right now, Isilon, uses an Infiniband back-end. Sure, Ethernet and fibre channel are fantastic transport, but Infiniband definitely has a place in the market. The extreme low latency and insanely high throughput with inbuilt fail over mechanisms makes for a damn good product.

Using this technology, Isilon is able to scale out cache and compute in a way that would probably be impossible with any other transport mechanism. It’s pretty amazing that you can grow an Isilon installation from a humble 3 node cluster all the way to 144 nodes without ever having the ‘old world storage’ concern for storage processor CPU utilization or a situation where you just don’t have enough cache.

I couldn’t mention Infiniband without bringing up another one of the coolest interconnect products on the market, Xsigo (http://www.xsigo.com/ – they now also converge over Ethernet, too). Where Xsigo is really cool is in its philosophy of abstraction of infrastructure. We’ve already abstracted compute, storage, and networking. CNA’s and UCS chassis are doing more to abstract the infrastructure, but Xsigo lets us do this to our older gear that isn’t ready for the refresh cycle. Or for situations where low cost servers as brute compute capacity. It is another one of those highly flexible solutions that makes the data center and infrastructure manager’s lives easier.

Thunderbolt is another one. Intel and Apple just released this new interconnect that has the potential to change how devices attach to our computers. It follows the very same theme which we see so much of in the enterprise space currently, convergence. Thunderbolt has some truly incredible potential thanks to  its 10Gbit/s throughput (with promise for 100Gbit/s down the road) and the fact that it is a direct extension of the PCI-Express bus. This is the first truly viable extension of the PCI-E bus outside of consumer-level devices. Thunderbolt could pave the way for some insanely cool technologies but there’s first a tough battle of market acceptance to be fought.

I spent some time recently discussing Thunderbolt with a colleague and in that discussion came up with some of my “dream technologies” which Thunderbolt could deliver with ease:

• External hi-end graphics processing for laptops. Why shouldn’t I be able to use any GPU on the market with my laptop by way of a simple external enclosure?
• Integration of (swappable) GPU’s into monitors. I think this is a more logical pairing of devices, imagine plugging your laptop (or any machine; thin clients anyone?) into a huge monitor that would offload the graphics workloads into the monitor in a way that is appropriate to drive that specific monitor.
• Even further, entire ‘computer accelerator’ boxes. With graphics, sound, PCI-E SSD, raid arrays, and port expansions. We could take a cheap netbook and with one cable turn it into a powerhouse.

Who gets to break the mold, though? If an engineer wont do it, then who?

The Architect.

I believe it to be incumbent upon all Architects to try to break out of convention in everything they do (if only ever so slightly). Certainly straying from convention cannot be to the risk of their customers, but I think it is critical that they are always considering novel and unique solutions to each problem. This Engineer/Architect paradigm absolutely is not confined to IT. Most obviously it persists into … wait for it .. architecture and construction. Engineers provide their value by their rigorous insight into standardized little boxes and Architects present the Engineers with new boxes to standardize and examine. In the end, each side has done the other a favor and the end result is often amazing if some attempt is made to try something new.

I want more of these technologies that offer new takes on the traditional paradigms of computing.

We, as humans, are composed, defined, and products of our individual and shared experiences. Each and every one of us are individually responsible for collecting and distilling them into how we see the world, who we are, and using them to guide our path through life. On our journey through space upon this little blue sphere we all have to make huge, big, scary, and individually life altering choices that define the course of experiences that we will encounter. More often than not, these choices mean we have to pack some of those experiences up in a small cardboard box and file them away. Maybe we will pull them out again later, maybe not. Sometimes, it’s just a box full of junk but, more often than not, that little neat package is worth far more than its weight in gold. Their value is measured entirely in the perspective they can offer to you for decades to come.

I recently had to make one of those big, scary, life altering decisions when leaving my last employer. But you better believe I have a box full of gold to bring along with me. Already I’ve found myself digging deep into those experiences. Technical knowledge, personal interactions, and people I’ve met have all come along with me. Though I am sure I’ve only seen the tip of the iceberg here, I already have collected some amazing experiences; and it’s all thanks to change.

These days, we’re seeing more change than ever. Everything moves and changes at the speed of light now and shows no sign of slowing its forward acceleration. For those of us who work in IT, we see this more than anyone. The whole game is changing with the IT buzzword of the decade: virtualization. It’s so much more than a buzzword for those who understand it and do not fear it. It is the great enabler of change.

At first, it was all about consolidation of workloads. But it is now about so much more. We are seeing evolution to facilitate changes to environments, without downtime, that no one could imagine only a year ago. By now, I think we are all familiar with the concepts of virtualized storage (SAN), virtualized CPU and Memory workloads (VMware ESX/VMware View), and virtualized and converged networking (VPN/CNA/FCOE).

Last Thursday, I had the opportunity to see one of the coolest new technologies on the market, in person. I was fortunate enough to attend a Varrow only demo of EMC’s new VPLEX product presented by one of the brightest technologists I’ve met, Scott Lowe.  Can you say “VM Teleportation”? I can, and can only imagine some of the amazing things it can provide to corporations. Despite all the abstractions that virtualization has afforded us, we have until recently always been at the mercy of a single physical location. There have been some solutions in the past that helped us get around this, but they’ve always been extremely complex and relatively high-touch. Never before have we been able to completely abstract the physical aspect of compute and storage away from that physical location. The days of workloads being tied to a physical location are coming to an end with VPLEX.

An increasing number are joining the 24x7x365 global computing environment and as such fewer and fewer are able to sustain a total datacenter outage even for minutes or seconds. Imagine being able to completely shut down one of your datacenters for maintenance. If your core applications and services are virtualized, VPLEX will allow you to do just that! Your virtual environment, through a little EMC and VMware magic, can move across town without any downtime and without anyone noticing and without anyone being forced to do some very heavy lifting to tote those cumbersome metal boxes that we call servers around.

I don’t think any of us will ever forget the painful task of night+holiday+weekend maintenance to move our environment to a new provider. Now, with the right gear, they can be something we only tell stories about. It will make all of our lives easier and help bring back more of that coveted and sometimes mythical work/life balance. But most of all, it can make the tough conversations with our bosses and customers about post-mortems and reason for outage reports a thing of the past.

Finally a quick message to anyone who reads this:
Feel free to contact me if you want to talk virtualization, storage, disaster recovery or business continuity. I’ve recently taken a role here at Varrow as a Technical Consultant and would love to come talk to you and help you solve some problems!

John Simmons – Varrow
Twitter: @InternetJohn
Email: jsimmons@varrow.com
Phone: 866-783-8604, x156