MOBILE CORE · COMPETITOR COMPARISON
Loosha MCL 1.19.1 vs Spirent Avalanche vs Spirent Landslide.
Mobile-core capabilities. Loosha MCL as of v1.19.0 (measured against open5gs); Spirent from public datasheets.
Protocols
| Capability | Loosha MCL 1.19.1 | Spirent Avalanche | Spirent Landslide |
|---|---|---|---|
| GTPv2-C (S11/S5/S8) | Yes: Create/Modify/Delete Session, Create/Update Bearer, Release Access Bearers, DDN (TS 29.274 Rel-19 IEs); live gtpv2-live on S11 (verified vs open5gs) and S5 (verified vs built-in fake PGW) | Basic only: SGW↔PGW Attach/Detach for tunnels (marked Phase II in datasheet) | Yes, full EPC |
| GTPv1-C (Gn/Gp) | Yes: Echo, Create/Update/Delete PDP Context (PCAP) | Yes: PDP Context Activation/Detach (SGSN↔GGSN) | Yes (2G/3G) |
| GTP-U (S1-U/N3/N9) | Yes: traffic profiles, Echo, Error Indication, End Marker; stateful TCP/HTTP inside GTP-U | Yes: L4-7 traffic inside GTP-U | Yes, hundreds of Gbps |
| PFCP (Sx/N4) | Yes: Association, Heartbeat, Establishment/Modification/Deletion, URR, mobility; live pfcp-live SMF→UPF with parallel workers (one association per worker) | No data in datasheets | Yes: N4, UPF isolation |
| NGAP (N2) | Yes, in PCAP (tshark-clean); since 1.19.0 live gNB emulation ngap-live over SCTP (NG Setup, registration, PDU session, Xn/N2 handover; Linux/WSL core, verified vs open5gs: 200/200 UEs at 50/s) | No | Yes: N2, eNB/gNB emulation |
| 5G NAS (N1) | Yes, in PCAP: Registration, PDU Session Establishment; since 1.19.0 live with 5G-AKA (Milenage) and NAS security 128-NIA2 / NEA0-NEA2 in ngap-live (SUCI null scheme only) | No | Yes: N1, 5G-AKA, EAP-AKA', SUPI encryption |
| IKEv2 / IPsec (NWu, N3IWF) | Yes, in PCAP since 1.18.0 (n3iwf): untrusted non-3GPP access with IKE_SA_INIT, IKE_AUTH (EAP-5G NAS), CREATE_CHILD_SA, INFORMATIONAL, GRE-in-ESP user plane (NULL ESP, Wireshark-decodable), optional NAT-T, N2 from the N3IWF and N3 to the UPF; analyzers pair IKEv2 exchanges and link IKE SAs to NGAP UEs; no live N3IWF emulation | No | Yes: 5G UE emulation over Wi-Fi towards the N3IWF and AMF (5G datasheet); 5G to ePDG handover |
| S1AP / EPS NAS | No | No | Yes |
| Diameter | Yes, generation: S6a (ULR/AIR/PUR/NOR), Gx, Gy | No data | Yes (S6a, Gx and others) |
| HTTP/2 SBI | Yes, in PCAP: HTTP/2 + HPACK in clear (h2c) or over real TLS 1.2/1.3 with an SSLKEYLOGFILE key log that tshark decrypts; Nsmf_PDUSession (N11), Npcf_SMPolicyControl (N7); no live NF emulation | No (HTTP/2 only as L7 traffic) | Yes: 20+ NFs (AUSF, UDM, PCF, NRF, SCP, SEPP...) |
| RADIUS / SIP / RTP | Yes: RADIUS Accounting, SIP/SDP, RTP/RTCP | Yes: SIP, RTP, HTTP, DNS etc. as L7 | Yes: VoLTE/VoNR |
Nodes
| Capability | Loosha MCL 1.19.1 | Spirent Avalanche | Spirent Landslide |
|---|---|---|---|
| UE | Subscriber as session record (IMSI, MSISDN, APN, IP); 5G NAS in PCAP only | L4-7 user inside the tunnel | Yes, millions of UEs |
| eNB / gNB | Since 1.19.0 live gNB emulator ngap-live (one source + one target gNB, N3 GTP-U echo); no gNB-CU/DU split | No | Yes, plus gNB-CU/DU (O-RAN) |
| MME / AMF | Live SGW-C/MME side of S11 (GTPv2); since 1.19.0 tests an AMF as DUT from the gNB side (ngap-live); no AMF emulation | No | Yes: test and emulation |
| SGW / PGW | Live S11 client verified vs open5gs; S5 verified only vs built-in fake PGW | Paired SGSN/GGSN, SGW/PGW emulation | Yes: SGW-C/U, PGW-C/U |
| SMF / UPF | Live SMF→UPF over PFCP; built-in fake UPF | No | Yes: SMF, UPF and CUPS topologies |
| Live exchange with DUT | Yes in Full edition, CLI only: pfcp-live, gtpv2-live, ngap-live (since 1.19.0, WSL), -workers, TCP/HTTP in GTP-U, probe; plus OS replay. Community edition: PCAP only (live commands exit with code 3) | Yes, stateful | Yes, fully stateful |
Scale and price
| Capability | Loosha MCL 1.19.1 | Spirent Avalanche | Spirent Landslide |
|---|---|---|---|
| Subscribers | Community 100 sessions, Full unlimited; tested live up to 100,000 PFCP sessions and 50,000 S11 sessions | > 100,000 UEs (GTP datasheet) | Millions of devices |
| Signalling rate | Measured vs open5gs on one PC: PFCP 100k sessions with 4 workers, 98,991 established at 1368 setups/s; GTPv2 S11 50k, 49,500 created at 494 setups/s. Against the built-in fake peers 13,000-28,000 setups/s, so open5gs is the bottleneck | > 40,000 tunnels/s | Tens of thousands of events/s |
| Latency | At 100k: PFCP Establishment p50/p99 41/65 ms; GTPv2 Create p50 143 ms (open5gs in WSL2, under load) | No data | Real-time KPI |
| Throughput | Software-only on a PC/NIC, no chassis, no hardware line rate | Up to hundreds of Gbps (C100, CF chassis) | Hundreds of Gbps |
| Hardware | Ordinary Windows PC (Linux build available upon request) | Spirent C100/CF chassis or virtual | C50/C100 chassis or virtual/cloud |
| Price | Community edition free for non-commercial use; Full Edition $89 one time | Unknown; expensive licenses on request | Unknown; expensive per-feature licenses (L-APP, L-FT) |
Procedures
| Capability | Loosha MCL 1.19.1 | Spirent Avalanche | Spirent Landslide |
|---|---|---|---|
| Attach / Create Session | Yes, PCAP and live (49,500 of 50,000 vs open5gs S11) | Yes, basic Attach | Yes |
| PDU Session (5G) | Full 5GC flow in one PCAP (5gc-e2e): N2 → N11 → N7 → N4 → N1N2 → N11 Update → N4 Modification; live: N4 (pfcp-live) and, since 1.19.0, N1/N2 PDU session establishment through a real AMF/SMF/UPF (ngap-live) | No | Yes, with slicing and SSC 1/2/3 |
| Handover | intra/inter-TA, inter-gNB, SGW-U/UPF relocation in GTPv2/PFCP; since 1.16.0 NGAP Xn (Path Switch) and N2 handover with two parallel N3 tunnels and End Marker, and 4G<->5G over N26 (GTPv2 Forward Relocation, Context Request/Response, Indirect Data Forwarding Tunnel), connected and idle (PCAP only) | No | Yes: X2/Xn, N2, 4G↔5G, EPS fallback, N26 |
| Idle / Paging | Yes: Release Access Bearers, DDN | No | Yes |
| Stateful L4-7 | Basic: TCP handshake, HTTP GET/200, FIN inside GTP-U; 500/500 flows vs open5gs (S11 and N4), 200/200 vs fake peers in 1.13.1 | Core strength: HTTP/S, video, VPN, connections/s | Yes, content emulation |
Negative, PCAP, reports
| Capability | Loosha MCL 1.19.1 | Spirent Avalanche | Spirent Landslide |
|---|---|---|---|
| Negative tests / fuzz | Yes: 3GPP negative cases for GTPv1/v2/PFCP, field-edit, impair, gtp-resilience; since 1.17.0 edge-cases (F-SEID collision, PFCP CH/CHID, EBI reuse, truncated optional IE, same-sequence retransmission), sequence-number edit and gtpu-orphan (late G-PDUs, Error Indication) | L4-7 attacks / security | Yes: negative AMF/SMF, SBI negative, chaos, SCAS |
| PCAP generation | Core strength: deterministic generation, merge, field-edit; tshark-clean | No data | Capture during tests |
| PCAP analysis and replay | analyze, 3GPP explain, correlate, reconstruct + profile scaling, replay; session analyzer pairs NGAP and GTPv2 N26 (Forward Relocation, Context) procedures and counts handovers (since 1.17.0); decodes IKEv2/ESP on NWu and links IKE SAs to NGAP UEs (since 1.18.0) | Traffic replay at scale (C100) | No data on replay |
| KPI reports | kpi-report from PCAP/PCAPNG and live runs, validated on real open5gs captures (100k PFCP, 593,841 packets; 50k S11): counts match tshark exactly. Reads loopback and Linux any captures; non-standard PFCP / GTPv2-C ports auto-detected (1.14.1), -pfcp-ports / -gtpv2-ports override | Real-time stats, reporting tool | Real-time KPI, CI/CD test packages |
| Automation | CLI (all 1.7-1.19 features). MCP server for AI agents (Claude Desktop, Cursor): analysis, KPI p50/p95/p99, live PFCP / GTPv2-C runs (pfcp_live, gtpv2_live, ngap_live; confirm=true, limits, Full license) and, since 1.15.0, every generator scenario from a subscriber/session file (generate_scenario: LTE auto-pick, required-field checks), verified vs open5gs. The GUI covers the PCAP tools and, since 1.15.0, scenarios from a session file | Tcl, REST API | REST/Tcl, automated packages |
| Learning / beginner aids | Since 1.18.1: bilingual (EN/RU) Interfaces and nodes guide in the manual (4G/5G map; 13 nodes, 15+ interfaces with protocol/port and the MCL scenario covering each), GUI hover tooltips with F1 / right-click What is this? help, Interfaces & nodes cheat-sheet poster; scenario_catalog lists the interfaces of each scenario | Product training and documentation (no data on in-tool aids) | Product training and documentation (no data on in-tool aids) |
| Certification / compliance | None | No data | SCAS testing listed |
Notes
- "No data" means the reviewed datasheets do not mention it; it does not prove the feature is absent. Loosha figures are from one PC against open5gs 2.7.6 in Docker/WSL2, not against a commercial DUT. Live modes in Loosha require the Full edition.