The easiest way to think of a Layer 1 (L1) switch, also known as a physical layer switch, is as an electronic patch panel. Completely transparent connections between ports is performed based on software commands sent to the L1 switch over its control interface. In testing environments, this allows the tests to be as accurate as if there were a patch cord between the devices.
Using Layer 2 (L2) switches for connection purposes in a test environment can cause a number of problems. In L2 switches, the output bit stream is different from the input bit stream in that MAC control frames, such as pause frames, are discarded by the MAC layer. Also the differing clock timing forces the PHY on the L2 switch to add/delete idle characters to compensate, making it impossible to compare input data streams to output data streams when testing using an L2 switch.
L1 switches on the other hand, are fully transparent to the traffic going through them. Once a connection is made between two ports the attached devices are essentially directly connected. L1 switches have very low latency and do not store or manipulate a single bit in the data stream. The internal hardware architecture of the L1 switch allows the L1 switch to duplicate any incoming data to any number of output ports at full wire speed without dropping a single bit. This allows testing of multiple devices from a single test set or output. Software simulation of cable breaks (port flapping) can also be simulated using software-defined duration times and repetitions of the simulated cable break.
Cold Fusion Chassis Options Chassis can be populated with any combination of Interface Blades making it the only L1 switch to support 10m to 128G in the same chassis
The 12 RU 8-Slot Chassis can be populated with up to eight Interface Blades. The Chassis also houses integrated fabric control components, variable-speed fan units that automatically adjust according to the system temperature, and hot-swappable AC power supplies.
Port capacities per 8-Slot Chassis vary depending on the type of Interface blade used. Maximum port count is:
– 256 any-to-any mapping ports of up to 128Gbps
– 1024 ports of up to 28Gbps/32Gbps Fibre Channel
– 512 any-to-any mapping ports up to 28G/32Gbps Fibre Channel – including RJ45
32 QSFP INTERFACE BLADE
Each 32 QSFP Interface Blade has 32 configurable QSFP28/SFP28 ports that support any MSA compliant QSFP module with data rates up to 128Gbps. Additionally, each port partitions into four 1Gbps- 28Gbps (32Gbps Fibre Channel) data lanes and using breakout cables each blade supports up to 128 data lanes.
64 SFP INTERFACE BLADE
Each 64 SFP Interface blade has 64 configurable SFP-based ports supporting any MSA compliant SFP module up to 28Gbps. (32Gbps Fibre Channel) Ethernet RJ45 SPF copper modules are also supported, and can populate the entire blade, for rates up to 10G.
L1 Testing Scenarios
Intermittent cable connections or fibre insertion/removal can cause the optical signal to fluctuate 100s of time before becoming stable. Link flaps can cause all sorts of network failures; from routing table recalculation to the PHY not being able to re-establish the link. The ability to identify and correct switch behavior under lab conditions can save hundreds of thousands of dollars in field failures and onsite support.
ColdFusion can imitate these conditions by shutting down the output optical signal for a user-defined number of mSec and a user-defined number of times. With precise timing and order of events, multiple links failures, short time between link up/down events and other field failures are easily simulated in the lab.
FIBRE CUT SIMULATION
Fibre cut simulation typically assumes that both fibres (both Tx and Rx) cut, but in real life, deployments, sometimes the fibre is cut in a way where only one of the strands in the cable is damaged. When this happens, losing one fibre out may not trigger the fail over mechanism because most fail overs rely on optical power monitoring.
ColdFusion simulates this event by mapping two ports in a unidirectional fashion and can be programmed to transmit in one direction, but not in the other i.e. Device B receives signal from Device A, but Device A does not receive signal from Device B eliminating the optical power from a single strand. This can test to make sure the fail over mechanism is triggered in these situations.
ColdFusion can also simulate a microfibre cut in 100G network applications by simulating the same scenario above; eliminating optical power to one of the ports of the 4×25 Gbps lanes in QSFP28 to insure that the device under test responds or triggers the proper fail over response.
ColdFusion is capable of perform multicast (1 to n) mappings. One transmit port can multicast a signal to any number of ports within the system with a user-settable return path to the test port. This allows one port of a very expensive test set to simultaneously generate a test signal to any number of ports in the switch.
ColdFusion maximises the use of your existing test ports, or can minimise the number of new ports required to meet your testing needs.
In the reverse of multicast mapping, ColdFusion can perform port mirroring which sends the same signal to any number of test devices for analysis. This 1 to n mirroring enables you to simultaneously run various tests on the same signal in a test configuration.
Cold Fusion’s port mirroring can impact the total amount of time needed to execute a test, freeing the equipment for the next test set up.
A test lab environment is filled with many different types of media. Think about the devices under test, test equipment, and any ancillary equipment used in the configuration of the test scheme and how you will establish the connectivity architecture. Some equipment only supports one type of media. Some media is less expensive to implement. Comparing the cost of single mode and multi mode transceivers at higher data rates, you’ll find single mode transceivers can be two to three times more expensive than its multi mode counterpart. Copper RJ45 connections can pose these same connectivity issues.
ColdFusion supports the largest range of media types. Additionally its OEO architecture allows conversion of single mode signals to multi mode, and copper to fibre, making mappings from a single mode port to a multi mode port or a copper port to a fibre port possible. Having this flexibility allows you to utilise equipment in a test configuration regardless of media type.