Replacing legacy SDH with MSTP or packet transport
Networks built on legacy SDH have three practical migration paths: extend the existing MSTP platform so it carries Ethernet alongside E1 and STM-N traffic, move to a packet transport network built on MPLS-TP and carrier Ethernet, or adopt a hybrid design where OTN/WDM transport carries both. YT CIG, a Shenzhen-based multi-vendor supplier since 2014, sources hardware for all three.
What Are Your Options for Replacing a Legacy SDH Network?
The three options differ mainly in how much of the installed base you keep. Option 1, extend MSTP: the SDH nodes already in service gain embedded Ethernet capability, so E1, STM-N and Ethernet all ride the same rings. This is the lowest-disruption path because protection, O&M procedures and the management system stay unchanged, but it inherits SDH’s rigid container granularity and reaches a ceiling when platform end-of-life arrives.
Option 2, packet transport: replace the circuit-switched core with MPLS-TP and carrier Ethernet, carried over an OTN or DWDM layer. This path uses bandwidth more efficiently for Ethernet-dominant traffic and adds native packet OAM, but E1 and other TDM services must be converted through circuit emulation (ITU-T Y.1453), and the whole timing plan changes.
Option 3, hybrid: an OTN/WDM underlay carries both the surviving TDM services and the growing packet services, while SDH islands stay at the edge until their traffic ends. This is the most common real-world outcome for multi-vendor estates because it decouples the two migrations.
All three paths preserve telecom-grade expectations: ring or linear protection in the tens of milliseconds and a coherent timing plan. The differences are cost pacing, skill requirements and how long E1-based services remain in the estate.
Three forces behind SDH replacement
SDH was engineered for voice and leased-line traffic, and its capacity ladder reflects that. ITU-T G.707 defines STM-1 at 155.52 Mbit/s, STM-4 at 622.08 Mbit/s, STM-16 at 2488.32 Mbit/s and STM-64 at 9953.28 Mbit/s, with each level exactly four times the previous one; STM-256 at 39813.12 Mbit/s is standardized but legacy metro estates in practice run rings between STM-1 and STM-64 (ITU-T G.707).
Three forces drive the replacement. First, the traffic mix: most revenue services are now Ethernet, and forcing Ethernet into fixed SDH containers wastes capacity when demand is bursty. Second, the lifecycle: many SDH product lines have passed their end-of-sale dates, so spares and expansion boards increasingly come from the refurbished market rather than factory channels. Third, operations: a packet transport platform carries native Ethernet OAM and statistics that SDH equipment cannot provide, which matters for SLA reporting on business services. SDH still works technically on all of these counts; what changes is the arithmetic of keeping a parallel circuit-switched layer for a shrinking share of traffic, and that arithmetic turns on the size of the remaining E1 tail.
How Does MSTP Extend the Life of an SDH Transport Network?
MSTP (multi-service transport platform) is the standard first step: an SDH transport platform with embedded Ethernet switching, standardized as Ethernet over SDH. The mapping stack is defined in ITU-T recommendations: GFP frames Ethernet packets for transport over byte-synchronous pipes in either framed (GFP-F) or transparent (GFP-T) mode (ITU-T G.7041); virtual concatenation combines multiple VC-12, VC-3 or VC-4 containers into one flexible pipe (ITU-T G.707); and LCAS adjusts the number of members in a virtual concatenation group without service disruption (ITU-T G.7042), with G.7043 extending the same mechanism to PDH rates (ITU-T G.7043).
For an operator with healthy SDH rings, this path needs no new fibre, no new topology and no retraining on packet protocols. Equipment in YT CIG’s catalog shows how far MSTP went before packet platforms took over: the Huawei OptiX OSN 3500 combines TDM and packet switching on one chassis, 200 Gbit/s TDM plus 160 Gbit/s packet across 16 service slots, with STM-1/4/16/64 interfaces and SDH, PDH, Ethernet, ATM and microwave services supported (Huawei OptiX OSN 3500 datasheet), and the ZTE ZXMP S385 delivers STM-16/STM-64 rates with a 256×256 VC4 cross-connection and roughly 35 Gbit/s of access capacity (ZTE ZXMP S385 datasheet).
The limit of this path is granularity and cost: a VC-4 pipe carries roughly 150 Mbit/s of useful payload per ITU-T G.707, so scaling Ethernet to 1G and 10G service levels inside SDH containers gets expensive per bit compared with native packet switching. MSTP works while the Ethernet load stays modest and the SDH estate is young enough to keep in service; past that point the per-bit arithmetic favours native packet switching.
The packet transport layer: MPLS-TP, OAM and protection
A packet transport network replaces the circuit-switched core with connection-oriented packet switching. MPLS-TP is the transport profile of MPLS: ITU-T G.8110.1 defines it as a layer network that forwards packets along statically provisioned LSPs set up through the management plane, and that static provisioning gives transport-style deterministic behaviour over an MPLS data plane (ITU-T G.8110.1). Carrier Ethernet rings complement it with Ethernet-layer protection.
Operationally, the packet layer brings its own OAM and protection standards. MPLS-TP OAM reuses the Y.1731 Ethernet OAM toolkit for fault management and performance monitoring (ITU-T G.8113.1); continuity-check messages sent at 3.33 ms intervals detect a failure within roughly 10 ms, leaving time for protection switching to complete inside the classic 50 ms window. Protection itself is specified in ITU-T G.8131 for linear 1+1 and 1:1 schemes and G.8132 for ring protection using wrapping or steering (ITU-T G.8131; ITU-T G.8132), while pure Ethernet rings use ERPS, which completes switching within 50 ms under the single-failure conditions defined in the recommendation (ITU-T G.8032).
Underneath, OTN provides the transport layer: OTU2 runs at 10709.225 Mbit/s and OTU4 at 111809.974 Mbit/s, which is what makes 100GE services transportable in one line rate (ITU-T G.709).
The practical differences from MSTP are three: bandwidth is switched as packets and LSPs instead of fixed containers, so per-bit cost drops at 1G/10G/100G service levels; OAM reports Ethernet-layer faults directly, where SDH infers them from VC alarms; and E1 traffic survives only through circuit emulation (ITU-T Y.1453), which adds a small latency and a new failure mode you must monitor.
Which Equipment Families Support Both SDH and Packet Transport?
Several platform families in current and secondary supply can carry both modes, which lets one estate run a single network through the transition. The table maps the families listed in the YT CIG equipment directory to verified datasheet capabilities.
| Platform | Vendor | Class | Verified capability (datasheet) | Role in migration |
|---|---|---|---|---|
| OptiX OSN 1800 II / V | Huawei | Box MS-OTN | Unified switching of OTN, packet and SDH; services from 2 Mbit/s to 100 Gbit/s; MPLS-TP support (Huawei OSN 1800 datasheet) | Hybrid access/aggregation node serving E1, STM-N and packet on one chassis |
| OptiX OSN 3500 | Huawei | Metro MSTP | STM-1/4/16/64; 200 Gbit/s TDM + 160 Gbit/s packet switching; MPLS/MPLS-TP packet and TDM dual role (Huawei OSN 3500 datasheet) | Existing metro rings upgraded in place toward packet |
| ZXMP S385 | ZTE | MSTP | STM-16/STM-64; 256×256 VC4 cross-connection; ~35 Gbit/s access capacity; SDH, Ethernet and ATM services (ZTE ZXMP S385 datasheet) | Keeps legacy ZTE rings alive while Ethernet services are added |
| OptiX OSN 9800 M12 / M24 | Huawei | OTN/WDM | Platform cross-connect 64 Tbit/s scalable to 100 Tbit/s; M24 supports up to 600G per channel with flexible grid; universal OTN/OSU/SDH switching (Huawei OSN 9800 datasheet) | Hybrid underlay consolidating TDM and packet on DWDM |
| 1830 PSS-16II | Alcatel-Lucent (now Nokia) | Packet-optical | Multiservice metro/regional transport platform (Nokia 1830 PSS datasheet) | Packet transport path for estates standardized on 1830 PSS |
The catalog also lists related families (OSN 500, OSN 2500, OSN 7500, OSN 9600 M12/U16 and OSN 8800) for operators extending a single-vendor lineage without changing platforms (see the full equipment directory).
How Does a Hybrid Migration Approach Work in Practice?
The hybrid design separates two decisions that operators often conflate: what carries the services, and what switching technology processes them. The transport layer is rebuilt first. A DWDM/OTN line system, for example the OSN 9800 family with its 64 Tbit/s scalable cross-connect (Huawei OSN 9800 datasheet), is laid across the metro or regional footprint, and every service, TDM or packet, is mapped into OTN frames per ITU-T G.709. Because OTN transparently carries SDH client signals, the existing STM-N rings can run over the new line system on day one without touching their protection logic.
The service layer migrates second. MS-OTN box platforms such as the OptiX OSN 1800 II/V perform unified switching of OTN, packet and SDH with service rates from 2 Mbit/s to 100 Gbit/s (Huawei OSN 1800 datasheet), so one chassis can terminate the legacy E1/STM-N tail on the TDM side and start aggregating MPLS-TP or carrier Ethernet on the packet side. As each SDH ring loses its E1 customers, its capacity is re-homed onto the packet layer of the same or an adjacent node, and eventually the SDH switching fabric in that node is simply no longer licensed or populated.
This staged approach also de-risks procurement: transport line systems and edge nodes are bought in separate waves, and refurbished units from the secondary market can cover the interim edge capacity while the packet build-out proceeds.
Which Migration Path Fits Which Network?
The right path follows from two questions: what share of your traffic is still TDM, and what is your timeline for 5G or data-centre services. The matrix below summarizes typical fits.
| Scenario | Recommended path | Rationale |
|---|---|---|
| E1-heavy leased-line tail (utilities, financial networks) | Keep/extend MSTP | TDM stays native; GFP/VCAT/LCAS add Ethernet without touching TDM protection (ITU-T G.7041; ITU-T G.7042) |
| Mixed metro with growing Ethernet share | Packet transport with MSTP interworking | MPLS-TP OAM and protection (ITU-T G.8113.1; ITU-T G.8131) carry business SLAs; E1 rides emulation |
| 5G mobile backhaul build | Packet transport with PTP timing | Cell phase sync of ±1.5 µs per 3GPP TS 38.104 needs G.8275.1 full timing support |
| Long-haul / DC interconnect consolidation | OTN/WDM underlay first | OTU4 at 111809.974 Mbit/s carries 100GE in one lambda (ITU-T G.709) |
| Multi-vendor estate, phased budget | Hybrid | Transport rebuilt once; switching technology migrates ring by ring |
Two cross-checks come before any commitment: count the E1s that have no stated retirement date, and ask each equipment vendor for the packet-upgrade ceiling of your installed chassis. If the E1 count is small and the ceiling is high, the packet path is safe to accelerate.
A seven-step migration sequence
The sequence runs transport first, then switching, then services, and each step produces the input the next one needs.
- Audit services and timing. Inventory every E1/T1, STM-N circuit and Ethernet service, and trace which nodes depend on SDH line timing for their clocks. This audit determines how long TDM support must remain and drives the timing design.
- Classify rings. Group SDH rings by traffic mix and remaining life: TDM-only rings, mixed rings, and already-packet rings. Each class gets a different retirement date.
- Build the transport underlay. Deploy the OTN/WDM line system first, because it carries both worlds. Verify OTN-level performance before any service moves.
- Upgrade nodes in place where possible. Platforms with dual TDM/packet switching, such as the OSN 3500’s 200 Gbit/s TDM plus 160 Gbit/s packet configuration (Huawei OSN 3500 datasheet), let one chassis change roles without a forklift.
- Cut over ring by ring, service by service. Move Ethernet services first (they benefit most from packet transport) then migrate or retire E1s via circuit emulation (ITU-T Y.1453).
- Verify before decommissioning. Use out-of-service Ethernet service activation testing per ITU-T Y.1564, which validates throughput, frame loss, latency and jitter against the configured CIR/EIR profile in a single test, and monitor in-service quality with Y.1731-based OAM (ITU-T Y.1564; ITU-T G.8113.1).
- Decommission and salvage. Retired SDH capacity moves into the resale and refurbishment channel; the secondary market for STM-16/STM-64 MSTP equipment remains active precisely because other operators are mid-migration.
What Timing and Protection Issues Must You Plan For?
Timing is the most underestimated migration issue. SDH networks distribute frequency naturally from line codes; when those line systems go, the packet network must deliver frequency and often phase. Two standard tools cover this: synchronous Ethernet, whose equipment clocks (EEC) are defined in ITU-T G.8262 on the basis of the SDH SEC clock parameters of ITU-T G.813, distributed under the architecture in ITU-T G.8261 (ITU-T G.8261; ITU-T G.8262); and PTP, whose telecom profile for full timing support (every network element participating) is specified in ITU-T G.8275.1 (ITU-T G.8275.1). The network-level budget for phase delivery is an absolute time error of ±1.1 µs at the network output, of which 1 µs is allocated toward the radio air interface (ITU-T G.8271.1). That budget exists to satisfy radio requirements: 3GPP TS 38.104 requires TDD cell phase synchronization within ±1.5 µs of a common reference for both FR1 and FR2, with tighter time alignment error of 65–260 ns for coordinated features between co-located antennas (3GPP TS 38.104).
Protection needs equal care. If the legacy design promises 50 ms recovery, verify that the replacement mechanism delivers it under your topology: G.8032 ERPS meets 50 ms under the single-failure conditions it defines, and the recommendation itself notes switching can take longer when multiple protection requests must negotiate (ITU-T G.8032). MPLS-TP networks hold the same window when OAM detection is fast: CV at 3.33 ms intervals detects faults within about 10 ms (ITU-T G.8113.1), leaving margin for G.8131/G.8132 switching (ITU-T G.8131). Re-test the full protection path of the replacement ring before the first customer circuit moves.
Frequently Asked Questions
Is E1 over MSTP the same thing as E1 over circuit emulation?
No. On MSTP the E1 keeps its native form: a VC-12 container mapped into the ring, protected by the existing SDH mechanisms and managed from the same system. Circuit emulation converts the E1 into pseudowires across an MPLS or MPLS-TP network and reconstructs it at the far end (ITU-T Y.1453), which adds a small fixed latency and creates a separate fault domain. The Ethernet side of an MSTP node is what needs the standards stack: GFP for framing (ITU-T G.7041), virtual concatenation for pipe size (ITU-T G.707), LCAS for hitless bandwidth changes (ITU-T G.7042), with platforms such as the Huawei OSN 3500 adding dedicated packet switching (200 Gbit/s TDM + 160 Gbit/s packet) when Ethernet load grows (Huawei OSN 3500 datasheet).
What happens to E1 leased-line customers after the SDH network is retired?
They move to circuit emulation: TDM signals are carried as pseudowires across the MPLS/MPLS-TP network per ITU-T Y.1453, then reconstructed as E1 at the far end (ITU-T Y.1453). Keep an inventory of emulation endpoints, because each one adds a small fixed latency and becomes a distinct fault domain to monitor.
Does MPLS-TP interwork with an existing IP/MPLS core?
MPLS-TP runs on the standard MPLS data plane, so forwarding is compatible; the difference is that transport LSPs are provisioned statically through the management plane under ITU-T G.8110.1, while an IP/MPLS core signals them dynamically (ITU-T G.8110.1). The handoff point needs an explicit LSP-stitching configuration; the static and dynamic sides do not join themselves.
Does the 50 ms protection promise survive a double failure?
Not on the same guarantee. G.8032 ERPS completes switching within 50 ms under the single-failure conditions defined in the recommendation, and the recommendation itself notes switching can take longer when multiple protection requests have to negotiate (ITU-T G.8032). MPLS-TP linear (1+1/1:1) and ring protection under ITU-T G.8131/G.8132 reach the same window when OAM detection is fast: CV at 3.33 ms intervals detects faults within roughly 10 ms, leaving margin for the switch (ITU-T G.8113.1). A topology that must ride through two simultaneous failures needs its own timing analysis.
What happens to timing when half a ring has moved to packet?
Frequency can still be traced through the remaining SDH nodes, because the synchronous Ethernet equipment clocks (EEC) of ITU-T G.8262 are specified on the basis of the SDH SEC clock parameters of ITU-T G.813, with distribution under the ITU-T G.8261 architecture (ITU-T G.8261; ITU-T G.8262). Phase is the part that needs its own plan: the G.8275.1 PTP profile with full timing support delivers it, inside the ±1.1 µs network output budget of ITU-T G.8271.1, of which 1 µs is allocated toward the radio air interface.
Can refurbished SDH or MSTP equipment still be sourced for interim migration needs?
Yes, the secondary market is active because operators are at different migration stages. YT CIG’s catalog lists 21,323 units across brands including HUAWEI, ZTE, CISCO, NOKIA, ERICSSON and JUNIPER (see the equipment directory), with condition (new, refurbished or used) confirmed on quotation and every unit inspected before dispatch, and warranty following the original manufacturer as confirmed on your quotation (support FAQ).
A related read: carrier Ethernet switches guide covers the aggregation switch layer that typically sits above a packet transport ring, and the DWDM transport platform page shows the line-system hardware used for the hybrid underlay.
Need a specific part?
Send the part number and quantity — we confirm condition, availability and price before you commit.