Stop Guessing, Start Testing
Testing protocols are a critical element of colo facilities
Colocation facilities make a lot of sense for companies that don’t want to own, operate, or manage an increasingly complex, specialized, and expensive data center (DC). Why not delegate the stress of keeping your data center running smoothly to a specialist? This means that for data center operators (DCOs), infrastructure reliability and emergency response capability is essential — downtime is not an option. In this environment, testing is a critical element in meeting customer service level agreements (SLAs) and/or internal performance objectives. In a modern data center, there are thousands of links, cables, transponders, and connections, or, to it put frankly: thousands of potential points of failure. The following is a list of common key test use cases for DCOs.
Use Case No. 1
DC-to-DC Interconnect (DCI) — Leased Connections
Due to the importance of the data stored in data centers, most DCOs back up data to another data center to ensure quick disaster recovery and safeguard their customers’ business continuance.
Operators need that DC-to-DC transmission to happen as quickly as possible, so they are linked by very high-capacity connections in various protocols. To maintain the integrity of these connections and to verify SLAs, DCOs need to perform tests on Ethernet line rates up to 100G and beyond, as well as optical transport network (OTN), coarse wavelength division multiplexing (CWDM), or dense wavelength division multiplexing (DWDM) circuits. Critical tests include RFC-2544 Benchmarking Methodology for Network Interconnect Devices, Y.1564 Ethernet Service Activation Test methodology; and RFC-6349 Framework for Transmission Control Protocol (TCP) Throughput Testing.
In addition, with such big and critical circuits connecting data centers, testing the underlying fiber integrity on those circuits is also prudent. All-in-one testers are highly recommended, as techs may use them throughout the life cycle of the network.
Use Case No. 2
Network Testing to Unmanned Data Centers
In a lot of cases, DCOs — especially hyperscalers — have automated and virtualized their operations to the point that, in some facilities, there are few (if any) humans to connect test gear and take measurements. Despite a lack of technicians on-site, the interconnections still need to be tested periodically, if not continually, to monitor key performance indicators (KPIs) and SLA commitments.
One solution is to install a rack-mounted test unit that can securely run high-performance bit error rate (BER) tests to remote, unmanned data centers. In remote locations, these units are securely accessed and controlled with the user’s choice of manufacturer-developed tools, like Viavi’s vNet Fusion platform for centralized test and service activation, or industry-standard methods, like standard commands for programmable instruments (SCPI) or virtual networking computing (VNC) with encryption. Critical tests include RFC-2544, Y.1564, and RFC-6349.
Use Case No. 3
With ever-growing data loads, DCOs and internet content providers (ICPs) need to send more traffic between data centers. To address that trend, many DCOs are using DP-16QAM modulation to create 200G wavelengths over their DWDM systems, thereby doubling the capacity over the same fiber. While beneficial, the danger is not testing the system before adding live traffic to it. There may be limitations on a particular wavelength that prevent it from achieving a 200G transmission rate — limitations that cannot be known without stress-testing the wavelength before putting it into service. Many DCOs don’t test these new 200G links simply because they don’t have the test capability.
One solution is to use a field tester with two 100G ports that can stress test a data center’s critical 200G WDM DCI link. BER tests performed should include QuickCheck, RFC-2544, and Y.1564.
Use Case No. 4
Customer-to-Data Center Connection
In a colo facility, every client has a customized solution. In some cases, a customer requires a private Ethernet circuit from their premises all the way to their rented rack/cage or to a particular service provider in the DC. When application problems arise, a customer will call the DCO first, and the DCO must be able to test the transmission quality on the customer’s circuit into the meet-me room as well as to the extended demarc as part of a standard SLA verification protocol.
In this situation, hand-held network testers let technicians quickly run several, standards-based tests on the circuit from a quick connectivity check to KPI verification regardless of the bandwidth. DCOs can quickly determine whether or not the customer circuit is the problem, which allows them to take the appropriate next step.Tests include RFC-2544, Y.1564, and RFC-6349.
Use Case No. 5
Dark Fiber DCI
Because of the criticality of DCI links, not to mention the expense, some DCOs choose to purchase or deploy their own dark fiber instead of leasing a managed service from a telco/MSO. However, dark fiber was often deployed to carry 10G services. Because 10G services are less sensitive, fiber certification/qualification tests, such as insertion loss, optic return loss, length, optical time-domain reflectometer (OTDR), dispersion (CD and PMD), attenuation profiles, etc., are not necessary. The challenge is that DCOs aren’t lighting dark fiber to put 10G on them anymore — they need to support 100G and more. If the dark fibers were not proofed for 100G, then service turn-up and performance could be problematic. Although 100G coherent transponders use polarization mode dispersion (PMD) compensation mechanisms, there are limits to what can be corrected/tolerated. Also, legacy fibers have more loss/attenuation variance from fiber to fiber, so a baseline is needed to ensure the fiber and specific wavelengths can support the high speeds needed for DCI. Once in service, and specific to high-speed reconfigurable optical add-drop multiplexer (ROADM) networks, optical signal to noise ratio (OSNR) needs to be validated, as it directly correlates to the throughput performance, or BER of a network link.
To ensure performance, technicians need to use a tester that enables bidirectional insertion loss, optical return loss, and OTDR testing from one optical port. Also required is an optical dispersion module to allow full fiber characterization of chromatic dispersion, PMD, and attenuation profile. Optical spectrum analyzers (OSA) then allow for the measurement of the OSNR.
Use Case No. 6
Intra Data Center BER Testing
A DC tenant wants to know about the transmission performance within the data center — from the point of presence (POP) to the rack, from the rack to another service provider on another floor, etc. The transmission quality is likely very high; however, the DCO has no way to provide the customer with a measurement report to verify the SLA.
Technicians need to measure network KPIs (latency/packet delay, frame loss, jitter/inter-packet delay variation, and burstability) down to the nanosecond for such short network links. Many testers now generate a report that can be printed or emailed immediately.
Don’t Wait Until It’s Too Late
In an environment tasked with supporting multiple tenants, the pressure to keep services running smoothly is intense. Proper testing tools ensure maximum uptime by enabling DCOs to quickly pinpoint and respond to any potential issues before the network suffers.