Storage Networks

Scott Shimomura

The Network Matters for Cloud Infrastructure

by Scott Shimomura ‎03-05-2014 01:53 PM - edited ‎03-26-2014 10:44 AM (4,363 Views)

What’s the first thing that comes to mind when it comes to networking and cloud? It probably isn’t Fibre Channel.

 

As far back as 2000, a various industry experts have been predicting the death of Fibre Channel. Sometimes it has been an industry luminary like Bob Metcalfe, inventor of Ethernet, who predicted in 2005 that Fibre Channel would be “over in ten years” (I guess I have less than a year to find a new job). In 2007, it was an “IT consultant” screaming for attention for another technology like iSCSI that tried and failed to predict the demise of Fibre Channel.  

 

Fast forward to 2013. After years of predictions that cloud technology will take over, there is a niche (InfiniBand) vendor proclaiming that “cloud data centres are being built without Fibre Channel” .

 

Now, we are a nearly a quarter into 2014 and Fibre Channel is alive and well, expected by industry analysts to grow modestly for the foreseeable future. Upon further review, the industry experts were wrong in 2000, Bob Metcalfe was wrong in 2005, the iSCSI proponents were wrong in 2007, and InfiniBand is still searching for an identity…

 

Yesterday,  Brocade and Rackspace announced the completion of the deployment of Brocade Gen 5 Fibre Channel SAN ... across its global network of data centers. Yes Rackspace, the global leader in hybrid cloud and founder of OpenStack, the open-source operating system for the cloud. In a world of seemingly infinite technology choices for cloud infrastructure, Rackspace chose Brocade Gen 5 Fibre Channel and EMC VMAX Storage to build out a reliable and simple storage infrastructure solution that would accommodate massive growth, dynamic scalability, and increasing efficiency requirements.

 

When we originally launched our Gen 5 Fibre Channel technology in 2011, we said the adoption of this technology would be based on the continued adoption of virtualization along with the emergence of cloud-based architectures and solid state storage. In 2014, Fibre Channel continues to be the protocol of choice for server and desktop virtualization;

Fibre Channel matters for cloud-based infrastructure and has become the de facto standard interconnect protocol for solid state storage arrays. Our predictions have proven to be more accurate than Nostradamus. Unfortunately, success in predicting technology trends hasn’t translated to success picking Powerball numbers.

 

Emulex’s Shaun Walsh, Senior Vice President of Marketing and Corporate Development, recently posted a video blog on their view of the state of Fibre Channel. I couldn’t agree more with his sentiment that the reports of the death of Fibre Channel have been greatly exaggerated. With continued adoption and a near-term industry roadmap for Gen 6 Fibre Channel and 128 Gbps speeds, the future continues to look bright for the Fibre Channel industry.

 

Follow Brocade on Twitter @BRCDcomm

Follow me on Twitter @scottshimo

Comments
by Scott Shimomura on ‎03-26-2014 10:33 AM

Is your network ready for tomorrow? http://www.youtube.com/watch?v=EEafQrqxPo0

 

by AJ Casamento on ‎03-28-2014 01:01 PM

Scott,



You make a number of excellent points. So  many executives have in their minds the concept that cloud infrastructure  means any application on any platform at any time for any reason with better than 90% utilization of the asset and it will cost them less. One  of the challenges that I see here includes the level of flexibility that is expected in that statement.  Especially with the level of uptime that's expected in the same environments.  If we can't know where the next workload is going to be moved or created  than is much as possible we have to build  infrastructure they can tolerate  those changes and accelerate deployment without downtime.


Humans have a number of good qualities, but patience is not one of them. So when we describe these environments and the amount of functionality that's going to be pushed to these networks  and the level of dependency the people will have on them  for their daily lives  the one thing I can promise you is that the network mus NEVER fail. The service level agreement (SLA)  that is provided by storage infrastructure to these environments is much more critical than most people realize .  Servers and applications can survive temporary outages in a network and generally speaking when the outage  is corrected the servers are still standing. However a similar outage in the storage connectivity will result in the question "what color is your favorite screen ?" And tolerance for this kind of outage will not be high.

 

The direction of the Brocade GEN5 platforms continues to provide  both the maximum uptime environment  that the storage  platforms require  as well as  continuing to achieve the management  level  that will give us granularity down to the virtual machine . in the end the individual application owner only cares about his application ,  not ESX  or any of the other applications  he happens to be sharing  infrastructure with . And so we need the visibility down to that single application.  Even more, when the application owner calls ,we must not answer the phone with  "me 10 minutes to figure out where you are the network and I'll call you back" .

 

So things like extending Clear Link to the servers and eventually to the storage and having the visibility of Flow Vision  to look at discrete problems  becomes priceless to the infrastructure . And that is what my mission-critical data center customers are telling me every day.

 

It's funny because I find myself thinking of that old commercial and I think it should say,

 

"If you haven't looked at Brocade lately,  then you haven't looked at Brocade."

 

 

ciao,

  aj

 

 

Labels