Friday, May 27, 2011

Re-Cabling Your Data Center Part #2

In today's post we will discuss the upgrading of your data center infrastructure.
OK you made the decision, you are going to upgrade the optical infrastructure of your data center, and it is time to start planning.  There is a lot to cover, knowing that in most instances the “D” word will be involved (DOWNTIME).  Staying focused on the proper goals is the best bet to stay on track.  So what should you be thinking about for this re-cabling job?  The answer to these questions can ensure you stay on track, so it is critical to think about the following points.

Design is the first important aspect.  Use the TIA-942 standard as a guideline.  I would highly recommend purchasing and researching this document.  It offers a roadmap for your infrastructure that will allow for growth and scalability.  A well designed infrastructure will allow hardware upgrades, even with currently unheard of connector types.  It will also make adds, moves and changes much easier.  The basic principle is to create a structured environment where patching can be done in the “front” of the equipment racks and/or cabinets, and then trunking cable is run from the equipment cabinets to an area referred to as the MDA or Main Distribution Area.  This is where all the patching can be done, without having to lift a floor tile or dig around in overhead conveyance!


Once you have the proper design that can last for generations of hardware, now you need to ensure the performance of the product you put in backs this up.  In other words, it will be pointless to do this re-cabling job if the products you install won’t support the next generation of hardware.  For an optical infrastructure the most important issue is insertion loss.  Insertion loss is the amount of light (measured in dB) that is lost from the transmitter.  Loss budgets shrink as data rates go up (insert chart).   Look for products that have low insertion loss.  Research what the standards are, who exceeds them and also how they are measured.  There are a lot of “smoke screen” tactics, such as using average insertion loss versus maximum insertion loss.  The average does not help you a bit, because you may get an assembly that is on the high side.  Ensure that the products maintain these insertion loss standards after multiple uses.  Demand documentation on these repeatability factors and insertion loss characteristics.

So you have a good design, and products that will last for the next few generations of hardware upgrades.  Now you need to make sure things will look nice and STAY looking nice.  Manageability; much of this is about planning ahead, and not allowing the “band aid” fix (the old 10 meter jumper patch for a 3ft distance).  It is easy to blame a tech for this, but did they really have what they needed?  Cables seem to become an afterthought.  Also examine the product sets themselves.  Will they work with your rack and cabinet infrastructure?  Do they align with other industry standards?  Are the product sets adding un-necessary cable bulk?

Lastly, respect layer one. Think about cabling when any changes are being planned for and also plan for changes on the fly.  Have a small inventory of jumpers; vary the lengths so you are prepared.  Are the product sets based on industry standard connections or are they proprietary?  You don’t want to end up in a position where you have only one source for a connector or interconnect point.  Also look at the warranty; does the warranty require a manufacturer specific “end to end” scenario, where if you use another manufacturer’s product your warranty is nullified?

If you keep these factors in mind, it should really help you to create a positive legacy.  You probably inherited a cabling mess, ensure that you left a cabling masterpiece!

Friday, May 20, 2011

Re-Cabling Your Data Center Part #1

In the first of a multi part post we will be exploring the key reasons data center managers should consider and how to implement a comprehensive approach when re-cabling their data centers.
Re-cabling the optical infrastructure in your data center is a daunting task. There are so many factors to think about, and that is just dealing with the physical cabling! There are other factors that must be negotiated. Typically this job will involve some amount of downtime for the data center. That adds on a lot of planning, politicking and action items. Best practice tells us to start at the beginning and move forward from there.

The "beginning" is examining WHY this may need to be done in the first place. If this is kept in mind, the goals of the project can be met and it really shouldn’t hurt that bad…


One of the major reasons the upgrade is needed is the rapid demand for more and more enterprise storage. What started with a few servers and possibly a mainframe a decade ago has now looks like a junkyard of hardware from various manufacturers that are tied together with an assortment of point to point fiber jumpers that an archeologist would have trouble separating.
Don’t take it too personally; most of us didn’t see this amount of growth coming. Enterprise storage has grown 30% year over year for the past 5 years! It’s not finished either; it is projected to grow at the same rate for the next 5 years as well! To handle all these storage needs, a cabling infrastructure must be modular and scalable. Running jumpers from point to point (directly connecting from one piece of hardware to another via duplex fiber optic patch cord) will simply not work.

Another reason is global IP traffic. This is expected to quadruple in the next four years. Video is one of the key factors here. People are visual, and video is so effective for so many business drivers (not to mention those funny or cute videos) that there is little chance of this demand shrinking. An optical infrastructure will need to handle the massive bandwidth required. Think of plumbing for this example. The typical fiber infrastructure will have a hundreds of “small pipes” running in every direction. These have come from years of smaller hardware upgrades where the old cabling could not be disturbed in fear of causing an outage. This has been the vicious cycle of the data center fiber optical infrastructure.


There are other issues that are prevalent as well.  Unreliable cabling is a big culprit for “random” downtime… These always tend to happen at the worst times. Fiber optic assemblies may all look very similar, but there are so many shortcuts that can be taking the manufacturing process that degrade the quality of the cable, yet are completely unseen to the naked eye…. Like ticking time bombs, just waiting to be unplugged and then plugged back in and. SNAP, broken fiber at 2PM on a Saturday right before your daughter’s birthday party…

There are more reasons, but these are the major and most common ones. The next topic to address will be what to think about when planning for this job.

Monday, April 11, 2011

Webinar: Hot Topics in Data Center Fiber Optic Infrastructure

If you didn’t have a chance to attend CABLExpress 's Wednesday webinar, “Hot Topics in Data Center Fiber Optic Infrastructure,” you can view the archived version below.

CABLExpress Senior Data Center Infrastructure Architect Rick Dallmann presented on OM3 and OM4 fiber cabling, Senior Product Manager (you might know him as Mr. Cable Talk) Josh Taylor explained parallel optics, and Engineering Manager Kent Goldsmith spoke about bend insensitive fiber.

To view more archived webinars, click here.

Thursday, March 31, 2011

Webinar: A Close Look at the MTP Connector

CABLExpress recently hosted a webinar about the MTP connector, which is at the forefront of 40/100 gigabit Ethernet fiber optic cabling revolution in the data center. This presentation took a deep dive into the MTP connector and how it relates to your cabling infrastructure.

View the webinar below, or view this and past webinars anytime here.


Monday, March 28, 2011

Earn BICSI CEU Credits!

Join CABLExpress for a webinar in April 6 and earn BICSI credits. In this webinar they will discuss parallel optics, OM3, OM4 and bend insensitive fiber and how all of it relates to the fiber optic structured cabling system in the modern data center. 

You will learn:
  • Why parallel optics is being used for 40/100G Ethernet transmission in the data center over multimode fiber
  • How parallel optics works
  • What OM3 and OM4 fiber means and how to determine which you should use
  • What bend insensitive fiber is and its value in the data center structured cabling system

 Webinar attendees can earn 1 Continuing Education Credit (CEC) for BICSIs RCDD, Installer, Technician or Certified Trainer designations.  Register today!

Monday, March 21, 2011

2011 AFCOM Data Center World

If you’re involved in the data center, you’ve heard of AFCOM’s Data Center World conference and expo. This spring, they’re taking over the Mirage Hotel and Convention Center in Las Vegas – and CABLExpress will be there. Find them near Lounge B at booth 1007; we’ll be talking to attendees about how they can Respect Layer One™ and optimize their data center infrastructure with our Skinny-Trunk fiber optic cabling solutions.

Download a free expo pass – we’ll see in you Las Vegas, March 27-31!    

Friday, March 18, 2011

Fiber Polarity Methods

In this edition of CABLExpress's Cable Talk, JT discusses TIA approved fiber polarity methods when using the MPO style (MTP®) connector in the data center. Light path, fiber cross, duplex patch cords, LC connectors, MPO modules and fiber optic trunking design all come into play.

Wednesday, February 16, 2011

Virtualizing Server Infrastructure & The Cabling Infrastructure




In this edition of CABLE Talk JT discusses the virtualization of the server platform and what that means to a fiber infrastructure in the data center.  A high performance and properly designed fiber infrastructure can ensure a smooth transition. Insertion loss, repeatability and manageability to avoid any downtime issues in the data center.

Friday, February 11, 2011

Ever Wonder - Parallel Optics

With the advent of the IEEE 802.3ba standard for 40/100G Ethernet, it is important for anyone in the optical structured cabling field within the data center to understand what parallel optics is, how it works, why it will be used and potential obstacles to overcome.   In a recent CABLE Talk Josh describes the Parallel optics in greater detail.

How it works: It works by sending a signal from the transceiver over multiple fibers instead of single fibers (using RX and TX respectively).

Why it will be used: Current optical technology does not allow from transmission much over 10G speeds using LOMMF (Laser optimized multi mode fiber).  It can be done with single mode fiber however the costs will be much higher.  Parallel optics offers a compromise of cost and use of potential existing data center infrastructure cabling.

Potential obstacles:
Use of the MTP®, in the data center this connector with either 12 or 24 fibers will need to be utilized.
Insertion loss, as data rates increase insertion loss budgets decrease.  Be mindful of your loss budget and ensure any cabling investments are researched with loss budget as critical criteria.

Tuesday, February 8, 2011

40/100 G - Parallel Optics



In this edition of CABLE Talk JT dives into parallel optics. With the recent ratification of the IEEE 802.3ba standard for 40/100 G Ethernet, parallel optics will be utilized for short reach data center applications.

Typically used in data center structured cabling environments running high speeds of 10G or even 40 or 100 Gigabit Ethernet, SAN (Storage Area Networking), Fiber Channel, FCOE (Fiber Channel Over Ethernet) with such manufacturers as Cisco, Brocade, EMC and others. Typical applications could be virtualization or internal cloud core data center applications.  MTP® is a registered trademark of US CONEC.