Between 02:00 and 05:00 on 12 September we replaced the aggregation switches at our Dhaka core and doubled upstream capacity from 20 Gbps to 40 Gbps.
The work was done in redundancy, meaning at no point was a single path in service. Subscribers on the affected segments saw at most one brief BGP reconvergence of around eight seconds.
Capacity is only half of the story. We also added a third international transit provider, which gives us commercial leverage and a genuine failover option. When one upstream has a routing incident, traffic now shifts automatically instead of degrading.
Peak-hour utilisation across the core is currently running at 38% of the new capacity. That headroom is deliberate: it absorbs evening peaks, large game updates and the seasonal spikes we see when school terms begin, without shaping anyone below their package speed.