The New Network
Explore Juniper’s vision for network innovation and how the company and industry are shaping the future with the new network
Showing results for 
Search instead for 
Do you mean 

Peeling Back the Onion

by Juniper Employee on ‎07-14-2011 11:42 AM

For most of the last four months I’ve been on the road, doing what I enjoy most: talking with customers about the benefits of a data center fabric and the QFabric architecture.  That is the best part of my job, and I haven’t had this much fun in a long time.


It’s been particularly interesting to observe what a mind-expanding concept the QFabric architecture is for most customers. They quickly grasp the benefits of a flat, any-to-any fabric to interconnect their data center infrastructure, and the concept of single-hop latency for all data flows while eliminating the challenges of managing the physical locality of processing and data is compelling as they virtualize their data centers.  The ability to manage the fabric as a single, logical entity while eliminating the need to run multi-link protocols such as Spanning Tree or TRILL is almost too good to be true.


Cisco’s latest “Fabric”: Really?

by Juniper Employee ‎06-28-2011 11:11 AM - edited ‎06-29-2011 08:48 AM

I know it’s been a while since I last had a chance to share my thoughts on this site, but I couldn’t resist this one.  Cisco recently announced its “End-to-End High-Performance Trading Fabric.”  A couple of things about this announcement struck me as particularly interesting. 

First, it doesn’t appear that there is any real news here other than the fact they released performance figures for their Nexus 3068 switch.  It is so unlike Cisco to publicize its performance data; low latency of its switches is not one of their strengths. Cisco has not published performance numbers on the Nexus 5548 and for a good reason. The 5548’s performance on packets sizes of 512B and less is egregious.


When is a Fabric not a Fabric?

by Juniper Employee on ‎04-05-2011 05:14 PM

On March 30, Cisco announced an update to their data center “vision.”  We are pleased to see that they have endorsed our view of the problem—that the legacy multi-tiered data center network architecture must change to enable a truly virtualized data center.  This evolution will begin with the architecture, where hierarchical tree structures will give way to flat fabrics.  We at Juniper have been promoting this vision for the last two years and Cisco has now joined the chorus.  Where we fundamentally differ is in our view of what constitutes a fabric and how one implements that fabric.  Whereas Cisco is attempting to build a flatter, non-blocking network using their existing and evolving Nexus switches, we have focused on building a fabric by rethinking the concept of the switch.


Taking the Network Out of the Network

by Juniper Employee ‎03-04-2011 09:53 AM - edited ‎03-06-2011 06:17 PM

Robert Juric made the following comment to one of my recent blog posts:


"You state, 'Thus the interfaces to servers and storage should track standards...', but why stop at the interfaces to the servers and storage?” Robert wrote. “I believe we should have standards throughout the entire network. I'm not saying the other vendors are doing a better job at this, FabricPath isn't standard-friendly either. At least with TRILL or SPB we have people working on defining a standard, a standard which can be implemented by any vendor in order to create an OPEN network fabric.”


This is an excellent question, and speaks to the heart of the design of the QFabric™ switch.  Now that we have gone public with the architecture, I can better respond.  Robert's question reflects his perception that the QFabric switch is a network and, like any network, he strongly believes that it should be built out of standard interfaces.  We at Juniper completely agree—networks should be open.  However this is where the confusion about the QFabric switch most often arises - the natural assumption that the QFabric switch is a network.  It is not.  The QFabric switch is, in fact, a switch.

Before the QFabric switch, if you wanted to scale a network beyond the capacity of a single switch, you were required to build a hierarchy of individual switches—the legacy tree structure we see in today’s data center networks.  This approach, which adds cost, latency, and operational complexity, has been adopted by every other vendor attempting to build flatter, non-blocking networks in the data center.



The Great Circle of Life

by Juniper Employee on ‎02-28-2011 03:12 PM - last edited on ‎05-09-2011 10:13 AM by Administrator Administrator

After 30 years in this industry, I am starting to detect a pattern.  Every so often, a new product comes to the market that forever changes the industry—and changes what is possible. This is a rare event; yet in my career, I have had the privilege to personally participate in four such episodes. 


The first time I was a hardware product manager at HP when we rolled out the world's first RISC microprocessor-based server.  It was seven times faster than a DEC VAX 11/780 at a quarter of the cost.  It was my job to convince customers to implement something called RISC into their data centers—ironically the most risk-averse part of the IT infrastructure.  More than once, I had to explain that "really, it is spelled with a C, not a K.”


At the time, RISC was a revolutionary new architecture based on the concept that less is more.  By simplifying the instruction set, it was possible to shrink more of the computer system onto a single piece of silicon. The result was dramatic.  Whereas the VAX implemented the system across 13 separate boards interconnected via a bus, we were able to implement the main parts of the server on a single board.  Faster, cheaper and more reliable, the child of Joel Birnbaum's powerful vision was a remarkable achievement.  It forever changed the way computers were built.  It was the logical path forward. 


It’s Here: Revealing the QFabric™ Architecture

by Juniper Employee ‎02-23-2011 10:02 AM - edited ‎02-23-2011 10:08 AM

At last the big day is here!  Juniper has been talking about the power of a data center fabric for two years, since we first publicly disclosed the existence of Project Stratus.  We’ve discussed why we were building a fabric and what it would mean to our customers.  But we have intentionally been silent on the specifics of the architecture and the underlying technological magic.


That changes today.  In simultaneous locations around the world, we are sharing the secrets of QFabric, the fruit of the Stratus project, via webcast; and introducing the first component of the fabric, the QFX3500 Switch.


A few hundred years ago, a Franciscan friar named William of Ockham professed a concept that today is known as Occam's Razor.  He stated that when confronted with multiple alternatives, the simplest path is usually the correct one.  That is the defining concept behind QFabric.  Pradeep Sindhu, our founder and spiritual leader, looked at the data center network and saw a simpler, more correct path to solve this most difficult of network problems.  We believe the path he chose can eventually transform every data center in the world.


It is funny how history repeats itself.  This pattern is particularly noticeable when one has spent more than 30 years in an industry that changes at a light speed…I’ve seen history repeat itself multiple times. 


35 years ago, when I was typing out my JCL punch cards for the mainframe, I needed to specify the parameters for managing the core memory used by my program.  Today, operating systems now fully manage memory in a server.  Essentially the need to actively manage memory has been abstracted by the O/S and is now transparent to the users.


A decade ago, when it came to managing the data on disks, we were required to run a volume manger to specify how data was striped across multiple spindles.  Then along came advanced filesystems such as ZFS, which now manage the volumes for us.  Like memory management, volume management has become transparent.


Server Virtualization and the Path to Enlightenment

by Juniper Employee ‎02-18-2011 08:26 AM - edited ‎02-18-2011 04:17 PM

The pace of change in the data center is brisk to say the least.  One of the most significant drivers of change is the broad adoption of server virtualization, which is designed to allow multiple applications to independently co-exist on the same physical server.  There have been many different approaches to server virtualization in the past:  “envelopes” in MVS (zOS); Mainframe Domain Facility from Amdahl; Dynamic System Domains and Containers from Sun; and so forth.  Today, the preferred solution is to use hypervisors to encapsulate applications and their operating system instances inside a virtual machine (VM).


It may seem like hypervisors such as VMware’s ESX have sprung out of nowhere.  In fact, the hypervisor has been more than 45 years in the making and can be traced back to a 1964 R&D project at IBM’s Cambridge research facility running on a modified IBM 360-40 mainframe.  Initially known as CP-40 and later as VM/CMS, it was eventually released as IBM’s first fully supported hypervisor in 1972 under the name VM/370.  Although it remained in the shadow of MVS, VM/370 proved to be the O/S that customers would not let IBM kill off. Today, it is known as z/VM and runs on IBM’s z-series mainframes.


So what part of Layer 3 don’t they get?

by Juniper Employee ‎02-14-2011 10:05 AM - edited ‎02-18-2011 04:16 PM

One of the keys in building a revolutionary new fabric architecture for the data center is not forcing unnecessary change on the rest of the data center.  Data centers want to evolve gracefully.  The goal is to unleash the promise of the modern data center without disrupting how the infrastructure connects to the fabric or how applications are implemented.


Thus the interfaces to servers and storage should track standards so that the existing servers and storage can connect while evolving to incorporate newer protocols such as Data Center Bridging (DCB), VEPA, and FCoE.  It should also be possible to implement existing applications on top of the fabric to take advantage of the lower latency and greater agility to improve the user experience, but without requiring any alterations to the applications themselves.  .


Fabric Computing: Gartner’s view for the future of the datacenter

by Juniper Employee on ‎02-07-2011 09:55 AM - last edited on ‎02-18-2011 04:15 PM by Administrator Administrator

I attended the Gartner Data Center conference in Las Vegas this past December.  George Weiss and Andy Butler, two Gartner server analysts I have known and respected for some time, presented a concept they call “Fabric Computing.”  They talked about a flexible fabric that could interconnect all the resources in the data center, all the way down to the processor cores, caches, and memory.  The goal is to create a very flexible, elastic infrastructure on which a variety of applications can be easily provisioned and efficiently operated.  This is the ideal infrastructure to enable cloud computing, as part of a private or public model.


Those who follow my posts know I have been talking about the power of fabrics in the data center for some time. Flat, any-to-any fabrics form the ideal network topology for the modern data center. I believe Gartner’s vision of fabric computing is on the verge of reality, albeit on a slightly modified basis.  Rather than one fabric, there will be two fabrics, dictated by the speed of light and electrons.


The Two Roles of the Data Center Network

by Juniper Employee on ‎02-02-2011 09:50 AM - last edited on ‎08-04-2011 04:40 PM by Administrator Administrator

For years, the primary role of the data center network was to connect users to applications.  Over time—even as these networks evolved from terminal networks (Bisync) to SNA, Token Ring, DECnet and, finally, to Ethernet—their role remained largely the same.  And because there was typically a human at the far end of the network, a certain amount of latency could be tolerated.


With the evolution to SOA-based applications and shared storage, however, the data center network has taken on a new role, becoming an extension of the server and its memory hierarchy. 


About The New Network

Exploring the vision for the networking industry and the issues shaping its future.

Subscribe to The New Network    RSS Icon

Our Bloggers

Rami Rahim
Chief Executive Officer

Profile | Subscribe

Pradeep Sindhu
Vice Chairman of the Board & CTO

Profile | Subscribe

Mike Marcellin
Chief Marketing Officer

Profile | Subscribe

Ankur Singla
Vice President of Engineering

Profile | Subscribe

Bob Dix
Vice President
Government Affairs &
Critical Infrastructure Protection

Profile | Subscribe

Juniper Networks Technical Books