Showing 0 products
Frequently Asked Questions
Parallel MPO or duplex wavelength-multiplexed?
Duplex where fibre already exists, parallel where new fibre is being installed or breakout is wanted.
Wavelength-multiplexed modules put all four lanes on one fibre pair using four wavelengths, and terminate in the same duplex connector as a 10 Gb link. They run over existing duplex plant, which usually makes them the right answer for a building backbone that is already installed.
Parallel modules use eight fibres through an MPO connector. That needs ribbon fibre and MPO patching, which is a different infrastructure - but it is what makes breakout possible, since each lane is physically separate and can be taken to a different destination.
So the decision usually follows the cabling: existing duplex fibre points to multiplexed modules, a new data centre build with MPO trunks points to parallel.
Parallel modules are typically cheaper for the same reach; multiplexed modules save eight-fibre infrastructure. Over a campus, the fibre usually decides.
What is MPO polarity and why does it cause trouble?
MPO connectors carry many fibres in one ferrule, and the transmit fibres at one end must land on the receive fibres at the other - which needs a consistent scheme through every component.
There are three defined polarity methods, distinguished by how the fibre positions map through trunk cables, patch leads and adaptors. Each is internally consistent; mixing them is not.
The failure is that a link does not come up, and nothing looks wrong - the fibres are continuous, the modules are correct, and the light is simply arriving at the wrong positions.
Make the choice at design time, document it, and use components from one system throughout. Mixing trunk cables and patch leads from different schemes is where estates get into difficulty.
Also watch the connector gender and the pin alignment - MPO connectors are pinned or unpinned and must mate correctly - and inspect the end faces, since an MPO ferrule holds many fibres and one contaminated position degrades one lane, which produces partial failure rather than a clean one.
How does breakout to four 10 Gb links work?
The switch presents the port's four lanes as four independent interfaces, and a splitter cable takes them to four separate 10 Gb connections.
On platforms that support it, the port is configured into breakout mode - usually requiring a reload or at least the port group to be reset - and appears as four sub-interfaces, each configured independently.
The cable is either a direct attach assembly with a QSFP+ end and four SFP+ ends, an active optical equivalent, or an MPO-to-duplex fan-out connecting to four separate optical modules.
The value is density and cost: one port on an aggregation switch serves four devices, at less than the cost of four ports.
The constraints are that not every port supports it, that support often comes in port groups so neighbouring ports are affected, and that all four links share one physical cable and one module - so redundancy designs should not put both halves of a resilient pair on the same breakout.
Is 40 Gb still worth deploying?
For new builds, generally no - 100 Gb using four 25 Gb lanes has displaced it, and 25 Gb has displaced it at the access layer. Its remaining role is extending existing estates.
40 Gb was a transitional step: four 10 Gb lanes, which meant more fibre and more module cost per gigabit than the 25 Gb lane generation that followed.
Where 40 Gb infrastructure exists it remains perfectly serviceable, and adding to it is often the sensible choice - the modules are inexpensive now and the cabling is in place.
For new installations, 100 Gb uplinks with 25 Gb access align better on cost per gigabit, on cabling, and on the breakout relationship between the two.
QSFP+ ports do retain value as breakout ports for 10 Gb connections, and a great many are used that way rather than for 40 Gb links.
The practical guidance: extend what exists, but do not design a new fabric around 40 Gb.
What cabling distances does QSFP+ support?
Around 100 to 150 metres on multimode parallel, 10 kilometres and beyond on single-mode - with the multimode figure depending heavily on the fibre grade.
Short-reach parallel modules over multimode ribbon reach roughly 100 metres on older grades and somewhat further on newer ones. That covers most in-building and row-to-row connections.
Wavelength-multiplexed modules over multimode duplex have their own figures, generally shorter than the equivalent 10 Gb link on the same fibre.
Single-mode variants reach 10 kilometres routinely, with longer-reach options available.
The grade of installed multimode matters more at 40 Gb than at 10 Gb, and a backbone that comfortably carried 10 Gb may not support 40 Gb over the same length. Check the installed fibre's grade and measured loss rather than assuming.
Where the existing multimode will not support it, the options are a single-mode module on a single-mode strand if one exists, or accepting a shorter run by moving the equipment.
Can QSFP+ modules be used in QSFP28 ports?
Usually yes - QSFP28 ports commonly accept QSFP+ modules and run at 40 Gb, which helps during migration.
The cage and mechanical format are the same, and most 100 Gb switch ports support 40 Gb operation with a QSFP+ module fitted. That allows a 100 Gb capable switch to connect to existing 40 Gb equipment without dedicated ports.
The speed is usually detected from the module or set by configuration.
The reverse is not true: a QSFP28 module in a QSFP+ port will not run at 100 Gb, because the port's electrical lanes are 10 Gb.
Check the specific platform, since a few high-density 100 Gb switches restrict which ports support the lower speed, and breakout support sometimes differs between the two module types on the same port.
For a phased upgrade, this compatibility is what allows the aggregation layer to be replaced first and the access layer afterwards.
How should MPO connectors be cleaned and inspected?
With MPO-specific tools, and inspected with a scope that shows every fibre position - because contamination on one position degrades one lane rather than failing the link.
An MPO ferrule holds eight or twelve fibres in a row. Standard single-fibre cleaners and inspection scopes do not work on it; MPO cassette cleaners and multifibre inspection scopes are required, and they are a necessary purchase for any estate using parallel optics.
The symptom of contamination on one position is a link that establishes with errors, or one that runs at reduced performance, because one of the four lanes is degraded while the others are fine. That is much harder to diagnose than a clean failure.
Inspect before every connection, clean if anything is visible, and re-inspect after cleaning - the standard discipline for fibre, but more important here because there is more to go wrong.
Keep dust caps fitted whenever a connector is not mated. An uncapped MPO in a cabinet collects contamination quickly.