Jerry Gilreath's blog

Shedding Light on Dark Fiber for Data Centers

When evaluating your data center connectivity, there are many reasons to consider dark fiber, including cost control, flexibility, security, and scalability. To quickly understanding the basics of fiber optics, see my blog posting, “Tech Primer: Dark Fiber, Lit and Wavelengths.

Importance of Dark Fiber for Data Centers

The number of internet connected devices known as the Internet of Things (IoT) is expected to reach 20+ billion by the year 2020, according to a recent Gartner report.

Gartner report - IoT growth from 2014 - 2020Likewise, cloud usage has been escalating at a similar rate, year over year. The reliance on cloud platforms such as Amazon’s AWS, Microsoft Azure, IBM SoftLayer and Google Cloud also continues to skyrocket, as indicated by cloud revenues seen in this report from Synergy Research Group.

These growth markets are driving enterprises and online businesses to a level of network dependence that is becoming hyper-critical.

Growth of Cloud Providers - Synergy Research Group Report

Connectivity is King

A loss of network connectivity or degraded network performance across a network connection can cause more than the loss of revenue. Poor network performance could even cause the loss of a life in the case of some environments like healthcare, public safety, and the military.

How vital is your network? When stability, along with latency, security, and bandwidth are at the forefront of the decision makers mind, then dark fiber may be the answer.

RagingWire understands that connectivity is of paramount value in a data center. As such, RagingWire has both partnered with connectivity providers and made a significant capital investment in telecommunications infrastructure to service our customer’s unique needs.

For example, in our Sacramento data center campus, we partner with multiple carriers to provide lit and dark fiber services that deliver excellent network performance of ~2ms latency to San Francisco, and ~4ms latency to the South Bay – a location jam-packed with cloud providers.

In our Ashburn, Virginia data center campus, we offer both lit and dark services to multiple carrier hotels and cloud locations, including AWS and Azure, providing sub-millisecond latency between your carrier, your data, and your data center.

In Garland, Texas, within the Dallas-Fort Worth Metroplex, RagingWire has built a fiber network that connects its 1,000,000 square foot of data center campus to over 128 locations in the Dallas and Fort Worth market, including popular carrier hotels and cloud providers.

The Good, the Bad, and the Ugly of Dark Fiber

Dark fiber may be the right decision for many of today’s infrastructure connectivity needs, but make sure you go into it with full awareness of its advantages and disadvantages.

The Good:

  • Cost-control: Dark fiber costs the same whether you intend to use 1Gb, 10Gb, or 100Gb.
  • Flexibility: You may run any protocol and any service. You may even choose to install your own multiplexing equipment and slice the fiber into multiple channels (generally 16 channels, but current off-the-shelf hardware allows for up to 160), each usable for 1Gb, 10Gb, or 100Gb.
  • Security: Public access telecommunications networks generally have multiple access points at various nodes throughout the network, whereas dark fiber routes are accessible only at each of the two endpoints of the fiber run.
  • Scalability: Service may be upgraded as required by simply using higher performance equipment. Available bandwidth on dark fiber is limited by only three things: Physics, current technology, and your budget.

The Bad:

  • Cost-control: When your bandwidth requirements are 1Gb or less, lit services will usually be less expensive than fiber when considering the initial lease of fiber and capital outlay for hardware. Additionally, long-distance dark fiber may be more expensive than purchasing a wavelength. You’ll have to do the math and figure out which meets your needs and budget.

The Ugly:

  • Reliability: Your architect will need to design around the fact that there is no built-in fault-tolerance or connectivity failure protection. This will usually require the purchase of a second diverse dark fiber path between your two locations.
  • Scalability and cost-control: Dark fiber is point-to-point. Unlike many other carrier products available, dark fiber does not allow for multiple end-points on a network. It may be necessary to purchase multiple fiber paths for larger networks.

Summary

When considering dark fiber from fiber providers instead of lit fiber or carrier services from telecom providers, it is beneficial to map your unique IT connectivity needs with the strengths and weaknesses of dark fiber. This mapping exercise should help shed some light on the best connectivity options for your custom environment.

Connectivity Questions

Is your data center carrier neutral? Carrier neutrality is vital when choosing a data center. You want your data center to freely allow interconnectivity between all carriers and other colocation providers. This protects your interests and allows for future scale, plus it maximizes your flexibility.

What types of lit connectivity are available? It is less important to focus on the number of carriers in the campus; instead focus on whether the carriers you care about are available. Also ask if their direct competitors are available. This will be helpful for bidding – to keep your primary carrier as cost competitive as possible.

Is dark fiber available? If so, where does it go? Does the data center have a dark fiber product or a partnership? Where does it go and is the pricing competitive? Does the data center have lit connectivity options or a partnership?

 

Tech Primer: Dark Fiber, Lit and Wavelengths

Some IT colleagues have asked me, “What is dark fiber and what’s the difference between lit and wavelengths?” Let’s begin by understanding the basic concepts of fiber optics and the difference between dark and lit fiber.

Difference between dark fiber and lit fiber

Unlike wire, which passes electricity through a metal conductor, fiber optic cables use a specialized glass or plastic allowing for data to be transmitted great distances by passing light through the glass. Fiber that isn't currently being used and has no light passing through it is called dark fiber.

Utilizing this fiber, telecommunications carriers can provide something called “wavelength” services, also known as “waves.” This works by splitting the light into various wavelength groups called colors or “lambdas”. Carriers sell these wavelengths to separate customers and then recombine the colors and transmit it across fiber. Therefore, lit fiber is fiber that has been lit with light by a carrier.

Dark and lit fiber explainedTo better understand lit fiber’s wavelengths, think of a rainbow where each color is a channel of light. Remember Mr. "ROY G. BIV" from middle school – Red, Orange, Yellow, Green, Blue, Indigo, and Violet?

Wavelengths essentially split a single fiber into channels. Unlike copper wire, which uses an electrical signal, fiber optic communications utilize either a LASER or a LED operating at a very high frequency. Fiber optic cables have the ability to carry much higher frequencies than copper cables. Traditional bandwidth throughput (1Gb/10Gb/40Gb/100Gb) will easily fit into a single color channel. Each fiber can be split into hundreds of colors, but a typical lit fiber is split into sixteen colors or lambdas.

The business of fiber optics

In the late 1990's, there was an uptick in the number of carriers building out dark fiber networks. In addition, there was a high degree of inter-carrier trades – a practice where carriers would swap dark fiber assets with other carriers in order to gain a foothold in markets where they were not present or had limited capacity. Inter-carrier trades coupled with mergers and acquisitions allowed even the smallest of carriers to offer competitive data transport agreements around the world.

However, a significant portion of this built capacity remained untapped for years. Carriers wanted to avoid potential long-term lost telecommunications revenues and were reluctant to enable competitors in their high margin wavelength services market. In addition, carriers did not want to cannibalize their often oversubscribed and lucrative Ethernet services market with inexpensive high-capacity fiber. For these reasons, many carriers today still do not sell dedicated fiber assets directly to customers.

New demand for bandwidth

Technology needs have changed over time. Enterprises have become more dependent upon cloud services, interconnected infrastructures have grown in number, and a massive growth in the Internet of Things (IoT) all require a large data communications infrastructure that can scale rapidly, predictably, and on demand.

To fulfill these needs, dark fiber providers have entered the market and are working to provide massive bandwidth, low latency, and high quality connectivity to the end customer in the form of raw glass: dark fiber.

For additional information on the pros and cons of dark fiber versus lit services from carriers, read my blog post titled, “Shedding Light on Dark Fiber for Data Centers.”

5 Ways to Fool-Proof Your Data Backup Strategy

World Backup Day was yesterday, March 31st.

Nearly three years ago, an aspiring Reddit.com user made a proposal to the world at large. His proposal was a modest one:

"I propose we have a "Back-Up Day", a day when everyone remembers to check that they have good backups of all their treasured data."

The proposal was met with a resoundingly positive reaction, and the day adopted was March 31st; ironically, the day before April Fool’s Day.

Since that time, World Backup Day has taken off like wildfire, and here we are – three years later, yet another World Backup Day in the books.

In honor of World Backup Day, I’d like to give you five points, of advice. These are by no means complete. They’re just common-sense notes from the perspective of someone that has been in the thick of it.

  1. Get it right the first time.
    Back in my consulting days, I had a favorite quote by Paul “Red” Adair, a famous oil well firefighter. I remember reading about his efforts after the Gulf War (1991) when he was brought in to extinguish the oil well fires left in the wake of the retreating Iraqi troops. Red, if I may be so familiar, said, “If you think it’s expensive to hire a professional to do the job, wait until you hire an amateur.”

    Sorry to say it, but sometimes you do get what you pay for, and when it’s the recovery of your business on the line, do you really want to skimp on your backup? Now, don’t get me wrong, you certainly won’t find me arguing against the case for affordability; I mean a company has to stay in business, right? There are certainly world-class providers of backup that are "affordably" priced – my advice is simple: be sure any company you choose backs up their product with solid and proven software and systems; they should staff round-the-clock professionals that understand and can operate the technology; make sure those people are augmented by the staff that engineered the solution; and lastly, make sure the product is supported, via an active support contract, by a manufacturer and vendor that is available 24x7.
     

  2. Trust but verify.
    If you’re like me, you hire competent people to run your systems, therefore I trust my people.  I buy top of the line systems and hardware, therefore I trust my systems. I buy enterprise class software, therefore I trust my software. I trust my people to make mistakes, I trust my hardware to fail when I need it most, and I trust my software to have bugs.

    Your vendor should be willing to discuss restore procedures with you, to do scheduled verification restores, and to exercise their software and hardware frequently. If you live in a world of compliance issues, your vendor should also be willing to provide logs and evidence of these actions.
     

  3. You don’t get to decide how important someone else’s data is.
    You heard me. Unless the data you’re protecting is your own, you don’t get to decide how important it is. There may be regulations and orders that specify retention periods, or availability – and these may meet your short term need, but until you have some tough conversations with department heads, you’ll never have a comprehensive backup strategy.

    Spend the time talking to the owners of your respective business units. Ask the important questions: What are your retention times; how many copies should be maintained online; what are your recovery time and recovery point objectives; what are common causes of data corruption, deletion, or destruction? Be sure to explain the jargon and educate at the same time – not everyone will be as savvy as you are.

    The business needs to understand that there are costs associated with meeting its need, but more importantly the business needs to understand that there is a need. Period.
     

  4. Tune and update your backups
    Just like networking, database administration, and email administration, Backup is a specialization. Backup systems are high performance environments that serve a niche purpose. Tuning and maintaining the environment that protects your assets shouldn’t come as an afterthought. Like any other specialized high performance system, your backup environment will perform differently under stress than under day to day operation. This includes the people that manage the environment.

    Pick an interval to evaluate your backup strategy and put it on your calendar. Ask yourself, have your systems changed and do you backup new systems? What happened to data from retired systems?  Are your backup policies still valid for the data? Are there any new retention orders (legal or otherwise) and can you produce evidence of backup success, failures, and verifications for audit purposes?

    Don’t forget your backup windows. Are your backup windows still comfortable? Are there things you can do to optimize your backups, such as segmenting data into multiple types and spreading the backup activities over the day and week, based on policy for each type, instead of midnight every night? And, if possible, leverage technologies like client side and server side de-duplication to reduce both long term storage requirements and the amount of time it takes to perform the actual backup.
     

  5. Reports and alerts should only go to those that will, and can, act on them.
    Sending reports and email alerts to someone who will just create an email rule or filter to delete it serves no purpose.  Instead, send success/fail/verify reports to the people that need to know their data is protected. Make the alerts actionable, and assign responsibility for ensuring error messages are addressed.

There you have it, five tidbits of advice – neither complete nor authoritative – simply serving as nothing more than a starting point for World Backup Day.

Remember, April Fool’s Day is today, the day after March 31st. If you’re like me, you don’t want to be caught without backups, because… well, that’s just foolish.

Subscribe to RSS - Jerry Gilreath's blog