Leased line speed test: how to check your BTnet speed

What to expect from a BTnet speed test

BTnet bandwidth is symmetrical and uncontended, so a healthy line should test at close to its port speed in both directions. Close, not exactly: BT's terms say the throughput you see in speed tests and applications will be lower than the port speed because of network and IP overheads, typically by 5-10%, and that the figure varies by application.

BT itself tests a new BTnet service over an Ethernet cable and does not connect or test wireless. That is a good model for your own testing.

How to test a leased line properly

1. Use a cable, not Wi-Fi. Wireless performance depends on distance, walls, interference and the device, so it rarely shows the line's full speed.
2. Test as close to the line as you can. Ideally plug the test device into a spare port on your router or firewall, so office switches and cabling are not in the way.
3. Use a capable device. The network adapter must be faster than the line: a gigabit adapter tops out around 1Gbps, so a 10Gbps line needs a 10Gbps adapter.
4. Test both directions. Upload matters as much as download on a symmetrical line.
5. Use a multi-stream test. A single connection often cannot fill a fast line on its own, especially at 1Gbps and above. Choose a test that opens several streams.
6. Try more than one test server. A busy or distant test server can be the bottleneck.
7. Repeat at different times, including a quiet moment when little else is using the line, and note the results with the time and date.

If you have a managed router or BTnet Meraki, plug into the LAN side of BT's equipment, which is where BT's service management boundary sits. On BTnet wires only, BT's boundary is the network terminating unit (NTU), so test through your own router or firewall and bear its limits in mind.

Why results can look low

Low results on a leased line are often caused by something on your side of the service boundary. Common causes:

• Wi-Fi: often the reason for low results; test on a cable
• Other traffic: backups, updates, video meetings or guest Wi-Fi using the line during the test
• Firewall limits: security inspection can reduce throughput well below a firewall's headline figure
• Device limits: an older laptop, a slow network adapter, or security software inspecting traffic
• Office cabling and switches that are not rated for the line's speed
• The test itself: single-stream tests, a congested test server, or a test site far from the UK
• Overheads: BT's terms put the normal reduction at typically 5-10%

Bandwidth Flex and resilience can also affect results. If your service has recently been flexed, check what the current port speed is. If you use load balancing, a single test may only use one of the two lines.

Latency and jitter

Speed is not the only measure. For calls and video, latency (the time data takes to make a round trip) and jitter (how much that time varies) matter as much as raw bandwidth.

BT's SLA includes a latency commitment: an average round trip of 20ms or less on its UK core network, between BT's selected point of presence and designated routers in the UK, and 95ms or less across its transatlantic link, averaged from 10-minute measurements over a calendar month. A speed test's ping figure measures something different: the round trip from your device to a test server, through your own network and across the wider internet. So use speed test latency as a quick health check, not as a measure of BT's commitment. See leased line latency.

Keep a baseline from day one

The most useful speed test is the one you run when the line is new. BT's terms give you five Business Days after BT tells you the service is ready to carry out your own acceptance tests; if you raise nothing in that time, the service is treated as accepted. Use that window well:

• Test at the edge using the method above, in both directions, at a quiet time and a busy time
• Check your key applications: a call, a video meeting, a large upload and access to your cloud apps
• Record the results with the date, time, device and test server, so you have a baseline
• Repeat after changes, such as a Bandwidth Flex, new firewall rules or a new office network

A baseline turns a vague feeling that the internet is slow into evidence. If later results drop well below it on a wired test at the edge, you know something has changed and where to start looking. See the installation process for where acceptance tests fit.

What to do if the results look wrong

Work from your side outwards:

1. Rule out the network. Re-test on a cable, at the router, with other traffic paused.
2. Check the equipment. Confirm your router, firewall and test device can handle the line's speed.
3. Record the evidence: times, results, test server and the device used.
4. Raise it. If tests at the edge still look wrong, tell us. We can help you work out whether the issue is on the line or in the network, and raise it with BT where it needs to go to BT.

Faults on BTnet are reported to BT's service desk under BT's support terms; see BT leased line support for how that works. BT's SLA covers availability, on-time delivery and latency; it does not set a throughput figure, so a slower test is a reason to investigate rather than an automatic SLA claim.

Not a customer yet and wondering whether a leased line would fix a slow connection? Contact us and we can help you size one.