CloudLabBox
Try the free demo

BGP practice labs across vendors

14 guided labs that take BGP from solid sessions to a small ISP edge on IOS, IOS XE, Junos, EOS and RouterOS: best-path selection step by step, one policy written in four languages, communities of every size, route reflectors, confederations, multihoming, RTBH and fast convergence, with two troubleshooting labs and a capstone built from a one-page design.

Start this courseTry it free for 30 minutes

Standard for a month, £75.00, runs every lab here, with nothing extra to pay for the guided labs. 12 of the 14 labs also run in the free demo.

Guided labs
14
Follows
BGP topics of Cisco ENARSI 300-410 and SPCOR 350-501, Juniper JNCIS-ENT and JNCIS-SP, Arista Data Center Engineering, MikroTik MTCINE and CCIE Enterprise Infrastructure
Runs every lab
Standard
Devices
IOSv, vEOS, RouterOS, vSRX NextGen template, Alpine Linux and CSR 1000v

Who these labs are for

For network engineers past CCNA level who run BGP or are about to, and anyone preparing for the BGP parts of ENARSI, SPCOR, JNCIS-ENT, JNCIS-SP, MTCINE or CCIE Enterprise Infrastructure.

The checks read the outcome: the path marked best, the attribute received, the hops a traceroute takes and what neighbours outside the AS see, not just whether a line of configuration exists.

The 14 labs

Every lab in the folder, as the workspace lists them, with what each runs and the smallest plan that fits. Read on 2 November 2026.

BGP in depth course 14 labs

Shared folderBGP in depth guided labs

BGP from sessions to path selection, policy and scale, then two troubleshooting labs and an ISP-edge capstone.

  1. 01BGP session basics on three vendors

    A partner on Arista and a customer on MikroTik are joining Kestrel Fibre's edge. Build sessions that survive a failed link by peering between loopbacks, then protect them with MD5 and TTL security, and see the same ideas in three vendors' commands.

    2× IOSv, 1× RouterOS, 1× vEOS

    4 vCPU, 3,328 MBFits LiteRuns in the free demo

  2. 02Best-path selection attribute by attribute

    Kestrel Fibre's change board wants every routing change predicted in writing before it goes in. Work down the BGP decision process one rung at a time, with a separate test prefix for each rung, and finish with one table that shows weight, local preference, AS-path length, origin, MED, eBGP over iBGP, IGP metric and router ID each deciding a real path.

    6× IOSv

    6 vCPU, 3,072 MBFits LiteRuns in the free demo

  3. 03One policy in four languages

    Kestrel Fibre has bought three small ISPs, and each runs a different vendor. Take one written routing policy and implement it as an IOS route map, a Junos policy statement, an EOS route map and a RouterOS filter chain, then prove each edge treats the same routes in the same way.

    2× IOSv, 1× RouterOS, 1× vEOS, 1× vSRX NextGen template

    6 vCPU, 7,424 MBFits LiteRuns in the free demo

  4. 04BGP communities of every size

    Kestrel Fibre lets customers steer their own routes with communities. Build the menu end to end: standard communities that set local preference, the well-known no-export and no-advertise, extended route-target tags, and large communities for a customer with a 4-byte AS number, which pass straight through routers that can't read them.

    3× IOSv, 1× RouterOS, 1× vEOS

    5 vCPU, 3,840 MBFits LiteRuns in the free demo

  5. 05Route reflectors and cluster design

    Kestrel Fibre's five core routers run a full iBGP mesh, and the sixth is on order. Replace the mesh with route reflectors, read the attributes that stop a reflected route from looping, and find out the hard way why two reflectors that share a cluster ID can leave a hole in your routing.

    5× IOSv, 1× vEOS

    6 vCPU, 4,608 MBFits LiteRuns in the free demo

  6. 06Confederations and their attribute rules

    Kestrel Fibre has bought Wharfe Telecom, and the two networks must look like one AS from outside without a flag-day iBGP redesign. Build a confederation across IOS and EOS, then prove which attributes survive a sub-AS boundary, how the AS path changes shape, and what one missing line does.

    5× IOSv, 1× vEOS

    6 vCPU, 4,608 MBFits LiteRuns in the free demo

  7. 07Multihoming and traffic engineering

    Tern Logistics connects to two ISPs and wants its web traffic in through one and its staff traffic in through the other. Control what goes out with local preference, then find out how far prepending and MED really get you on the way in, and why a provider's community is sometimes the only tool that works.

    4× IOSv, 1× RouterOS

    5 vCPU, 2,304 MBFits LiteRuns in the free demo

  8. 08Stability and safety nets

    One customer leaks a table, another flaps a link all night, and a partner restarts its routing software. Put the safety nets in place on Kestrel Fibre's edge: prefix limits, route-flap dampening, soft reconfiguration and route refresh, and graceful restart between IOS and EOS, then trigger each failure and read what happens.

    2× IOSv, 1× RouterOS, 1× vEOS

    4 vCPU, 3,328 MBFits LiteRuns in the free demo

  9. 09MP-BGP for IPv6

    Kestrel Fibre is turning on IPv6 for two customers. Bring up IPv6 address families on IOS, EOS and RouterOS, fall into the classic trap of IPv6 routes over an IPv4 session, fix the next hop, filter what customers send, and prove it with a ping between two hosts.

    2× Alpine Linux, 2× IOSv, 1× RouterOS, 1× vEOS

    6 vCPU, 4,352 MBFits LiteRuns in the free demo

  10. 10Edge hygiene with filters and RTBH

    Kestrel Fibre's edge accepts anything and passes everything on. Filter bogons and over-long prefixes, reject private AS numbers and absurd paths, stop a customer's hijack, stop being a free transit between providers, then stop a DDoS with remotely triggered black holes, at your own edge and at your provider's.

    6× IOSv

    6 vCPU, 3,072 MBFits LiteRuns in the free demo

  11. 11Fast convergence on IOS XE

    A route reflector shows each client only one path, so when an exit fails, Kestrel Fibre waits for BGP to find another. Give the IOS XE edge a second path with add-path, pre-install it with BGP PIC, detect failures in a second with BFD, then pull a link and watch.

    4× IOSv, 1× CSR 1000v

    5 vCPU, 6,144 MBFits LiteRuns in the free demo

  12. 12Troubleshooting BGP sessions

    Monday morning at Kestrel Fibre: five BGP sessions are down after a weekend of maintenance, and each one is stuck or failing in a different way. Use the session state, the logs and the TCP details to find five faults, including a session that comes up, then dies, every 15 seconds.

    4× IOSv, 1× RouterOS, 1× vEOS

    6 vCPU, 4,352 MBFits LiteRuns in the free demo

  13. 13Troubleshooting path selection and policy

    Every session is up and the graphs look fine, yet customers are complaining. Six policy faults across IOS, Junos, EOS and RouterOS are hiding in Kestrel Fibre's network. None of them shows up as a down session. Work from symptoms to causes with each vendor's tools, and prove each fix with the path it actually chooses.

    3× IOSv, 1× RouterOS, 1× vEOS, 1× vSRX NextGen template

    7 vCPU, 7,936 MBFits Standard

  14. 14Capstone - a small ISP edge

    Your last job at Kestrel Fibre: build the whole BGP edge from a one-page design, with no steps to follow. A route reflector on IOS, a transit edge on Junos, a peering edge on EOS and a customer edge on RouterOS, all tagging routes on the way in, preferring customers over peers over transit, and sending only what they should.

    4× IOSv, 1× RouterOS, 1× vEOS, 1× vSRX NextGen template

    8 vCPU, 8,448 MBFits Standard

Devices and images

The exact Add Node images the labs use, and the vCPU and RAM their nodes are set to. Your plan counts what each running node is set to.

Devices and images in the BGP in depth practice labs, as read on 2 November 2026
DeviceImage in Add NodeSet toLabsNodes
Cisco IOSvvios-adventerprisek9-m.spa.159-3.m91 vCPU, 512 MB1452
Arista vEOSveos-4.33.1.1F1 vCPU, 2,048 MB1010
MikroTik RouterOSmikrotik-7.14.21 vCPU, 256 MB99
Juniper vSRX NextGen templatevsrxng-23.4R2-S3.92 vCPU, 4,096 MB33
Alpine Linuxclbalpinebase-alpine-base-3-20-31 vCPU, 512 MB12
Cisco CSR 1000vcsr1000vng-universalk9.17.03.08a-serial1 vCPU, 4,096 MB11

Browse every image in the Add Node list

How to open the labs

  1. Sign in to your workspace at lab.cloudlabbox.com.

  2. In File Manager, open Shared, then the folder BGP in depth guided labs.

  3. Open a lab and start its nodes, then open Lab Tasks for the steps.

The labs are read-only. You can start the nodes and configure the devices, but changes to the topology can’t be saved. To change a lab, rebuild it in your own folder.

The Shared folder in the EVE-NG File Manager, recorded on 2 November 2026 with a test account.
The EVE-NG File Manager open on the read-only Shared folder: one folder per guided course, including Arista EOS, BGP in depth, CCIE Enterprise Infrastructure, CCNA, CCNP ENARSI and ENCOR, Check Point, Cisco ASA, Cisco Nexus NX-OS, CompTIA Network Plus and FortiGate
A guided CCNA lab with its Lab Tasks pane, recorded on 1 November 2026 with a test account.
A guided CCNA lab open in EVE-NG with its nodes running: four routers and two PCs on the canvas, and the Lab Tasks pane beside them with the scenario, the addressing plan and the first task

Questions about the BGP in depth labs

Are these BGP in depth labs free?

They’re included with every lab plan, with nothing extra to pay, rather than free. Plans start at £3 a day or £0.50 an hour. To try one first, the free demo gives you 30 minutes with 6 vCPU and 8 GB, enough to run 12 of these labs.

Which plan do I need for these labs?

Standard, with 12 vCPU and 16 GB, runs every lab here. Lite runs 12 of the 14. EVE-NG admits a node by the vCPU and RAM configured on it, so a lab needs the total its running nodes are set to. The biggest here, “Capstone - a small ISP edge”, needs 8 vCPU and 8,448 MB. Each lab’s Lab Tasks pane names the plan it needs.

Can I change these labs?

You can start the nodes and configure the devices, but the labs are read-only: changes to the topology can’t be saved. To change one, rebuild it in your own folder.

Are these labs from the vendors?

No. CloudLabBox wrote them to follow the topics of Cisco’s ENARSI 300-410 (v1.1) and SPCOR 350-501 (v1.1), Juniper’s JNCIS-ENT (JN0-352) and JNCIS-SP (JN0-364), Arista’s Data Center Engineering, MikroTik’s MTCINE and Cisco’s CCIE Enterprise Infrastructure (v1.1), where they cover BGP. They aren’t training material from the vendors, and CloudLabBox is independent of Arista Networks, Cisco, Juniper Networks and MikroTik.

Do I need vendor licences for these labs?

No. The vSRX nodes use the image’s built-in 60-day trial, and the other images need none for these labs. CloudLabBox doesn’t provide vendor licences or subscriptions.

Which plan do I need?

Twelve of the 14 labs fit Lite. Lab 13 and the capstone need Standard, because they add a vSRX to a larger network.

Why do some labs avoid certain vendors?

Because the images differ. IOS 15.9 has no large communities, so lab 04 reads them on EOS and RouterOS, and RouterOS 7.14 has no confederations, so no MikroTik takes part in lab 06. The panes say so where it matters.

Try a lab before you choose

The free demo is a real workspace for 30 minutes, with 6 vCPU and 8 GB: enough to run 12 of these labs. No card and no email.

Try it free for 30 minutesStart this course

Plans from £3 a day or £0.50 an hour, each with every guided lab.

More practice labs

CloudLabBox is independent of Arista Networks, Cisco, Juniper Networks and MikroTik. The labs follow exam topics and aren’t vendor training material; vendor names identify the exams and the images in the workspace.