You are here: / / / 2018

Dual-Power Feeds in Data Centers

Blog Category: by

Things like always-on technology, streaming content and cloud adoption are creating high demand for efficient, resilient and fast data centers that never let us down.

To meet these needs, dual-power feeds – two independent electrical feeds coming into a data center from the utility company – are becoming more common to reduce the chance of a complete outage (or not having enough power). This type of power set-up is often seen in Tier 4 data centers. If one of the two power sources suffers from an interruption, the other source will still supply power.

Generally labeled “A” and “B” feeds, each power source has not only its own utility feed, but also:

  • A backup generator
  • A switch that alternates between A and B feeds
  • Electrical and distribution switchboards
  • An uninterruptible power supply (UPS)
  • A power distribution unit (PDU)
  • Rack-level PDUs

At any one of these points along the chain, failure can occur. A true dual-power feed means that there are two separate sets of these components operating independently, reducing the likelihood of downtime due to failure.

Today, most mission-critical IT equipment, such as servers and switches, are also designed with at least dual power supplies. When everything is running normally, the equipment pulls power equally from both power feeds. In the event of an outage, however, the IT equipment can automatically switch all power to one feed or the other.

Read full article

Network Upgrades: Utilizing Parallel Fiber Cabling

Blog Category: , by

It comes to no surprise, that enterprise and consumer demands are impacting data centers and networks. As speed requirements go up, layer 0 (the physical media for data transmission) becomes increasingly critical to ensure link quality.

Numerous organizations are looking for an economical, futureproof migration path toward 100G (and beyond). Multimode fiber (MMF) cabling systems continue to be the most popular, futureproof cabling and connectivity solution.

Both duplex and parallel cabling are options for network upgrades. A few weeks ago, we discussed duplex MMF cabling. In this, we’ll discuss parallel MMF cabling.

 

Parallel Fiber Cabling

When transceiver technology can’t keep up with Ethernet speed requirements, the most obvious solution is to move from duplex to parallel fiber cabling.

Although using BiDi (bi-directional) and SWDM (shortwave wavelength division multiplexing) transceivers can reduce direct point-to-point cabling costs, they do not support breakout configuration (e.g. 40G switch ports to four 10G server ports), which is a very common use in data centers.

According to research firm LightCounting, approximately 50% of 40GBASE-SR4 QSFP+ form factors are deployed for breakout configuration; the other 50% are deployed for direct switch-to-switch links.

As a matter of fact, 40G QSFP+ and 100G QSFP28 are the most popular form factors used for Ethernet switches in data centers. QSFP (quad small form-factor) is a bi-directional, hot-pluggable module mainly designed for datacom applications. QSFP+/QSFP28 has a 2.5x data density compared to SFP+/SFP28, using four parallel electrical lanes. The optical interface is a receptacle for MPO female connectors. Four fibers (1, 2, 3 and 4) transmit the signal; the other four fibers (9, 10, 11 and 12) receive the optical signal.

QSFP transceivers, paired with parallel fiber connectivity with a one-row MPO-12 (Base-8 or Base-12) interface, can support flexible breakout or direct connection.

  • 40G/100G direct links are typically used in switch-to-switch links, which can be supported by duplex or parallel fiber cabling.
  • 40G/100G Ethernet ports can be configured as 4x 10G or 4x 25G ports to support 10G/25G server uplinks.
  • 40G/100GBASE-SR4 transceivers only use eight fiber threads in an MPO-12 connector; therefore, Base-8 is a cost-optimized cabling solution that allows 100% fiber utilization.

Read full article

Analyzing Data Center Energy Consumption By Using Business Metrics

Blog Category: by

About five years ago, the industry first heard about Digital Service Efficiency (DSE) – a method that was designed by eBay to help the company capture a holistic picture of their data center energy consumption and performance.

The initiative was then made public in an effort to assist other organizations establish their own data center energy consumption benchmarks and goals, and compare live system performance against those benchmarks and goals to determine actual efficiency levels.

While they tracking their data center’s power usage effectiveness (PUE), which illustrates how efficient a data center’s electrical and mechanical systems are, they felt like something was missing. Calculating PUE didn’t offer them insight into how efficiently their data center equipment (such as servers) was being used. The DSE initiative was formed to fill this gap.

Earlier this year, the team of eBay engineers who created the DSE initiative received a patent for it. With this news, we thought it would be a good time to revisit the data center productivity metric they introduced a few years ago. Even though it was created based on eBay’s core competency – e-commerce – there are still some lessons to be learned.

In eBay’s case, to measure performance and data center energy consumption, they chose to specifically measure how many online business transactions are completed per kilowatt-hour consumed. They calculated this by analyzing four metrics:

  1. The type of performance they wanted to measure (transactions, or the number of online purchases and sales)
  2. Cost per transaction (they measured cost per megawatt-hour, per user and per server)
  3. Environmental impact (amount of carbon dioxide produced per transaction)
  4. Revenue per transaction (they measured revenue per transaction, per megawatt-hour and per user)

Then they base their data center improvement goals around those metrics – goals like reducing cost per transaction by a certain percentage, for example, or increasing transactions per kilowatt-hour by a certain percentage.

The organization believes that, by substituting your own unique business metric in place of the metric they used – online business transactions – you’ll be able to create your own, unique way of measuring data center productivity and efficiency, too.

What performance metric could you use to measure and benchmark data center energy consumption? Here are a few ideas:

  • Healthcare: number of patients seen or number of appointments set
  • Hospitality: number of guests who stay onsite or number of reservations
  • Manufacturing: number of widgets produced
  • Financial: number of transactions

Read full article

Public vs Private Clouds: How Do You Choose?

Blog Category: , , by

An Intel Security survey of 2,000+ IT professionals last year revealed several fascinating information about public and private cloud adoption. For starters, within the next 15 months, 80% of all IT budgets will have some income dedicated to cloud solutions.

Many enterprises are starting to rely on public and private clouds for a few simple reasons:

  • Most good public and private cloud providers regularly and automatically back up data they store so it is recoverable if an incident occurs.
  • Tasks like software upgrades and server equipment maintenance become the responsibility of the cloud provider.
  • Scalability is virtually unlimited; you can grow rapidly to meet business needs, and then scale back just as quickly if that need no longer exists.
  • Upfront costs are lower, since cloud computing eliminates the capital expenses associated with investing in your own space, hardware and software.

But before you decide you are moving to the cloud, you should know the differences between public and private clouds. Making a choice between public and private clouds often depends on the type of data you’re creating, storing and working with.

 

Public Clouds Defined

The public cloud got its kick start by hosting applications online – today, however, it has evolved to include infrastructure, data storage, etc. Most people do not  realise that they have been benefitting from the public cloud for years (before most of us even referred to “public and private clouds”). For example, any time you access your online banking tool or login to your Gmail account, you’re using the public cloud.

In a public cloud, data center infrastructure and physical resources are shared by many different enterprises, but owned and operated by a third-party services provider (the cloud provider). Your company’s data is hosted on the same hardware as the data from other companies. The services and infrastructure are accessible online. This allows you to quickly scale resources up and down to meet demand. As opposed to a private cloud, public cloud infrastructure costs are based on usage. When dealing with the public cloud, the user/customer typically has no control (and very limited visibility) regarding where and how services are hosted.

 

Private Clouds Defined

In a private cloud, infrastructure is either hosted at your own onsite data center or in an environment that that can guarantee 100% privacy (through a multi-tenant data center or a private cloud provider). In these third-party environments, the components of a private cloud (computing, storage and networking hardware, for example) are all dedicated solely to your organization so you can customize them for what you need. In some cases, you’ll even have choices about what type of hardware is used. No other organization’s data will be hosted using the equipment you use.

With an internal private cloud (one hosted at your own data center), your enterprise incurs the capital and operating costs associated with establishing and maintaining it. Many of the benefits listed earlier about choosing cloud services don’t apply to internal private clouds, especially since you serve as your own private cloud provider.

In organizations and industries that require strict security and data privacy, private clouds usually fit the bill because applications can be hosted in an environment where resources aren’t shared with others; this allows higher levels of data security and control as compared to the public cloud.

 

What’s a Hybrid Cloud?

Enterprises also have the opportunity to take advantage of both the public and private cloud by implementing a hybrid cloud, which combines the two.

For example, the public cloud can be used for things like web-based email and calendaring, while the private cloud can be used for sensitive data.

Read full article

Network Cables; How Cable Temperature Impacts Cable Reach

Blog Category: , , , by

There is nothing more disheartening than making a big investment in something that promises to deliver what you require – only to find out once it is too late that it is not performing according to expectations. What happened? Is the product not adequate? Or is it not being utilised correctly?

Cable Performance Expectations

This scenario holds true with category cable investments as well. A cable that can not fulfil its 100 m channel reach (even though it is marketed as a 100 m cable) can derail network projects, increase costs, cause unplanned downtime and call for lots of troubleshooting (especially if the problem is not obvious right away).

High cable temperatures are sometimes to blame for cables that don’t perform up to the promised 100 m. Cables are rated to transmit data over a certain distance up to a certain temperature. When the cable heats up beyond that point, resistance and insertion loss increase; as a result, the channel reach of the cable often needs to be de-rated in order to perform as needed to transmit data.

Many factors cause cable temperatures to rise:

  • Cables installed above operational network equipment
  • Power being transmitted through bundled cabling
  • Uncontrolled ambient temperatures
  • Using the wrong category cabling for the job
  • Routing of cables near sources of heat

In Power over Ethernet (PoE) cables – which are becoming increasingly popular to support digital buildings and IoT – as power levels increase, so does the current level running through the cable. The amount of heat generated within the cable increases as well. Bundling makes temperatures rise even more; the heat generated by the current passing through the inner cables can’t escape. As temperatures rise, so does cable insertion loss, as pictured below.

Testing the Impacts of Cable Temperature on Reach

To assess this theory, I created a model to test temperature characteristics of different cables. Each cable was placed in an environmental chamber to measure insertion loss with cable temperature change. Data was generated for each cable; changes in insertion loss were recorded as the temperature changed.

The information gathered from these tests was combined with connector and patch cord insertion loss levels in the model below to determine the maximum length that a typical channel could reach while maintaining compliance with channel insertion loss.

This model represents a full 100 m channel with 10 m of patch cords and an initial permanent link length of 90 m. I assumed that the connectors and patch cords were in a controlled environment (at room temperature, and insertion loss is always the same). Permanent links were assumed to be at a higher temperature of 60 degrees C (the same assumption used in ANSI/TIA TSB-184-A, where the ambient temperature is 45 degrees C and temperature rise due to PoE current and cable bundling is 15 degrees C).

Using the data from these tests, I was able to reach the full 100 m length with Belden’s 10GXS, a Category 6A cable. I then modeled Category 6 and Category 5e cables from Belden at that temperature, and wasn’t able to reach the full 100 m. Why? Because the insertion loss of the cable at this temperature exceeded the insertion loss performance requirement.

Read full article

Easy, Cost-Effective Way to Add Power with Industrial PoE Injectors

Blog Category: , by

PoE Injectors can appease the growing power demands of energy-hungry devices in applications like physical security, transportation and automation – all in one device.

  • High-efficiency, low-waste power
  • Plug-and-play installation
  • Up to 240W of power from 8 ports

For recently developed or retrofit applications in need of maximum power without device limitations, these Power over Ethernet (PoE) injectors supply a high port count and up to 240 W of power.

PoE injectors join Hirschmann’s family of products built with industrial-grade housings and specific features to provide reliable power for industrial applications. They are the easiest and most cost-effective way to add high PoE power to both new and existing applications.

Benefits

  • Choose between active (integrated power supply) or passive (standalone module) devices for increased flexibility, depending on your needs.
  • Supports up to 240 W across 8 ports without load sharing, ensuring maximum power output. Each port can provide the maximum output power of 30 W.
  • Simple plug-and-play capability and compact size saves time and space while automatically detecting connected devices.

Features

  • Benefit from up to 8 available ports that deliver 30 W of power each
  • Enable PoE communication with a high number of devices using just one PoE Injector
  • Save costs with an all-in-one-solution and an efficient transfer of power (less wasted power) of >95 percent
  • Use in extreme environmental conditions, including wide temperature ranges (-45 °C to +85 °C for injector, -25 °C to +70 °C for injector plus power supply)
  • Install quickly and easily with automatic device detection and classification (IEEE 802.3at)
  • Meet important industry standards
    – Safety of Industrial Control Equipment: EN 60950-1, EN 61131-2, UL 60950
    – Transportation: EN 50121-4

Download Bulletin

Read full article

10 Factors to consider when Choosing a Rack PDU

Blog Category: , by

In it’s simplicity, rack power distribution units (PDUs) are designed to provide electrical protection and distribute power to networking equipment within racks/cabinets. As the needs and requirements of data centers altar, so do options for rack PDU performance.

There are several questions to consider before selecting rack PDUs that will work well for your data center application. This list below will aid you in the right direction, ensuring that the PDUs you choose will fit the design of your data center today and in the future.

1. Type of Mount

Depending on where you want to station it, a rack PDU can be mounted horizontally or vertically. Installed horizontally inside the rack (taking up RU space) is one option; another option is to vertically mount a PDU on the back or side of the enclosure (not taking up any RU space). You will often see one vertically mounted PDU on the left side and one on the right side of a data center cabinet (although rack PDUs can be mounted on either side, based on preferences).

PDUs can be mounted so that power cords exit either at the bottom or top of the enclosure. (If your data center is on a slab, for example, the power cord needs to exit at the top of the enclosure because there is no raised floor for it to pass through.)

2. Amperage

Your power rating – the amount of sustained power draw a PDU can handle – determines the amperage level you’ll need. Why is this important? Because, for example, a PDU with a 30A fuse will blow if a 30A circuit experiences more than 30A of power for an extended period of time.

Per the National Electrical Code, 30A PDUs or higher are required to be equipped with a 20A breaker to prevent injury in the event of a short circuit.

3. Voltage

In addition to different amperages, there are different input voltage options for rack PDUs as well; 208/240V is the most common voltage output to computing gear, with a new trend moving toward 400V input. Confirm your infrastructure voltage, and you’ll know what type of voltage you need in your PDU.

4. Single- or 3-Phase Power

What type of input power do you have access to: single-phase power or 3-phase power? The type of power distribution in your data center will determine whether you need a single- or 3-phase PDU.

The difference involves where in the distribution system the phase is broken down. When it’s broken down at the distribution panel, power to the rack will be single-phase service (requiring single-phase rack PDUs). When all three phases are brought to each rack, then a 3-phase PDU is needed. In most data centers, the input power is 3-phase service.

Read full article

Which is Right for You: 40G vs 100G Ethernet?

Blog Category: , by

Companies like as Google, Amazon, Microsoft and Facebook started their migration toward 100G in 2015 – and smaller enterprise data centers are now following suit. Plenty of these new 100G deployments adopt a singlemode fiber solution for longer reach that best suits their hyperscale data center architectures.

Comparing 40G vs. 100G optical transceivers currently available in the market, both have been developed and cost optimized for their designated reach and applications.

While weighing 40G vs. 100G Ethernet, and deciding which migration path makes more sense for your organization, here are some facts you should know:

  • Switches with 10G SFP+ ports, or 40G (4x 10G) QSFP+ ports, can support 10G server uplinks
  • Switches with 25G SFP28 ports, or 100G (4x 25G) QSFP28 ports, can support 25G server uplinks
  • 100G switches have already been massively deployed in cloud data centers; the cost difference between 40G vs. 100G is small
  • Most new 100G transceivers can easily support 40G operation
  • Some non-standard 100G singlemode transceivers are designed and optimized for cloud data center deployment; product availability for other environments is limited for the short term
  • Traditional Ethernet networking equipment giants Cisco and Arista have already started selling switch software on a standalone basis that goes into networking devices (such as a “white box” solution with merchant switch ASICs); this move accelerates hardware and software disaggregation and lowers overall ownership costs for end-users
  • According to Dell’Oro, 100G switch port shipments will surpass 40G switch port shipments in 2018.

When considering system upgrades from 10G, it’s essential to understand that 40G will also be needed to support the legacy installed base with 10G ports; 40G/100G switch port configurability will certainly accelerate 100G adoption in the enterprise market.

In 2017, 100G Ethernet is already ubiquitous – it will be mainstream, not just in hyperscale cloud data centers. Next-wave 200G/400G Ethernet will soon hit the market; standards bodies have already initiated a study group for 800G and 1.6T Ethernet to support bandwidth requirements beyond 2020.

Wrapping Up the Road to 800G

We’re almost finished with our blog series covering the road to 800G Ethernet. Subscribe to our blog to follow this series, as well as receive our other content each week. As part of this blog series, we’ve covered the following topics:

 

Read full article

Expectation for Fiber Connectivity: Layer 0

Blog Category: , by

The footprints of cloud data centers continue to increase substantially to accommodate massive amounts of servers and switches. To support sustainable business growth, many Web 2.0 companies, such as Google, Facebook and Microsoft, have decided to deploy 100G Ethernet using single mode optics-based infrastructure in their new data centers.

According to LightCounting and Dell’Oro, 100G transceiver module and switch port shipments this year will outpace last year’s shipments, with 10 times as many being shipped in 2017 vs. 2016. Shipment for 200G/400G switch ports will begin in 2018.

Data Center Architecture and Interconnects

Most intra-rack fiber connectivity has been implemented with DAC (direct-attach cables). As we discussed in our fiber infrastructure deployment blog series, system interconnects with a reach longer than 5 m must use more fiber connectivity to achieve the desired bandwidth.

100G, 200G, and 400G transceivers for data center applications have already been showcased by various vendors; massive deployment is expected to start in 2018. Based on reach requirements, different multimode and signal optical transceivers are being developed with optimized balance between performance and cost. Examples include:

  • In-room or in-row interconnects with multimode optics or active optical cables (AOCs), with a reach of up to 100 m. (New multimode transceivers, such as 100G-eSR4, paired with OM4/OM5 multimode fiber, can support a maximum reach of up to 300 m for 100G connectivity, which is suitable for most intra-rack interconnects.)
  • On-campus interconnects (inside the data center facility), with transceiver types such as PSM4 (parallel singlemode four-channel fiber) or CWDM4/CLR4 (coarse wavelength division multiplexing over duplex singlemode fiber pair) for 500 m reach.
  • On-campus interconnects (between data center buildings), with transceiver types such as PSM4 and CWDM4/CLR4 for a reach of 2 km.
  • Regional data center cluster interconnects, also referred as data center interconnects (DCIs), using coherent optics (CFP2-ACO and CFP2-DCO) for a reach of over 100 km, or direct modulation modules, such as QSFP28 DWDM ColorZ, for reach of up to 80 km.

Multimode Fiber Roadmap to 400G and Beyond

Multimode optics use low-cost VCSELs as the light source. When compared to singlemode transceivers, which utilize silicon photonics, VCSELs have some native performance disadvantages:

  • Fewer available wavelengths for wavelength division multiplexing
  • Speed is limited by the singlemode laser
  • Less advanced modulation options
  • High fiber counts needed to deliver required bandwidth
  • Shorter reach in multimode fiber (limited by fiber loss and dispersion) compared to singlemode fiber

Read full article

IP-Based Systems and PoE in Digital Buildings

Blog Category: by

Digital buildings, smart buildings, intelligent buildings, connected buildings – no matter what you name them, the sentiment is the same: A building with devices and systems that are designed to collect and share data to run as efficiently as possible without human intervention.

IP-based systems – also known as networked systems – are what make this idea possible. These systems use Internet Protocol (IP) to communicate with each other through IP addresses and data packets. All types of building devices can be IP-based:

  • Access control
  • AV systems
  • Building controls/HVAC
  • Digital signage
  • Fire/life safety systems
  • LED lighting
  • Surveillance cameras
  • Voice/data systems
  • Wireless access points (WAPs)

To function, an IP-based system needs access to power and data. When deployed in digital buildings, they offer many benefits:

Simple Scalability

Only need 15 surveillance cameras today? Then that’s all you need to install. If you decide you need more devices, the system can quickly and easily be expanded. If you decide that you need fewer devices, they are easy to uninstall. The system doesn’t require you to install a certain number at a time.

Read full article

PRODUCTS

CUSTOMER SERVICE

WHERE TO FIND US

JHB: +27 11 444 1039

Unit 3, 7 On Mastiff, Mastiff Road, Linbro Business Park, Sandton, Johannesburg, South Africa

CPT: +27 21 447 4247

Block C, Podium Level, The Boulevard Office Park, Searle Street, Woodstock, Cape Town, South Africa

Copyright © 2024 Jaycor International
Engineered by: NJIN Agency