The Internet broke for ViewQwest broadband customers on 31 May 2025. This comes after another severe disruption that begun on 30 April 2025, lasting a few days for some affected customers. I wasn’t affected the last time, but I am this time around, and hence the need to vent some frustration here.
Internet access is a critical utility. In this day and age, it’s not just for tech-savvy folks who need to be connected, but pretty much something that everyone needs, and whose quality of life will be impacted by the loss of such utility. Internet service providers, along with all those involved in enabling the Internet to function, need to understand the importance of the service they are providing. To this end, it is incumbent upon these businesses to ensure they are staffed by competent people who are capable of managing all aspects of their critical operations professionally.
ViewQwest said of their 31 May disruption that it was caused by a power outage at an Equinix data centre. I am left very disappointed to hear a disruption in one data centre can bring down ViewQwest’s operation, so severely that not just their customers got impacted, but their very own official website became unavailable.
I wonder if ViewQwest felt any embarrassment at all that their website could be impacted by a power failure at just one single location. They should seriously do some soul-searching there.
A quick note for those unfamiliar with network and data centre operations: A data centre power disruption may seem like a rather disastrous situation. It is, definitely so, at the data centre level. Networks, however, are supposed to be designed so that a simple outage at the equipment, rack, or even data centre should not impact overall service. For IT infrastructure designers, this is something very simple to plan for by having redundant failover systems in different locations.
That said power outage was at Equinix SG1 data centre, and it lasted from 11:09 am to 12:24 pm. I have been monitoring a variety of network metrics from my home, including pinging to Google’s 8.8.8.8 DNS. The start of the SG1 power disruption matches exactly when I began to lose reachability to Google’s 8.8.8.8 DNS.
You would have expected that Internet service providers would have redundant equipment, including at multiple sites, so that their services can operate continuously without any interruption despite a variety of disaster scenarios. In principle, the outage of one site should have had at most just a few seconds of outage.
In my case, it appears that reachability to Google DNS, for me, took 5 minutes to be resolved. Perhaps this is alright to some people, but as a network-trained person, this looks to me like redundant systems that were supposed to failover automatically did not quite work as planned. It seemed like it took some manual intervention by engineers at a NOC (network operations centre) to restore service.
I know for some customers, the problem continued beyond those 5 minutes. I was lucky that the Internet came back sooner, but others did not. They could not even ping their default gateway, i.e. the router on ViewQwest’s side, which seems to suggest that ViewQwest had no redundancy there.
Very interestingly, while Equinix SG1 was said to have restored power at 12:24 pm, this restoration of power must have caused more trouble for ViewQwest. This is because, for me, reachability to Google’s 8.8.8.8 DNS was lost again. At that time, I was at home to be able to do some troubleshooting. It appeared that some parts of the Internet were working, but not everything. Services hosted in my home PC was also not reachable from at least some parts of the Internet.
I found this quite odd. It looked like ViewQwest staff must have tried something earlier, so that Internet was normalised for some of their customers, but when Equinix SG1 restored power, it broke more things in ViewQwest’s network. So the mild annoyance from a brief interruption earlier has now become frustration over ViewQwest’s inability to provide reliable network service.
Worse yet, it seems they could not resolve whatever problems they were having quickly enough. It took over 1 hour, before Internet access was restored at about 1:40 pm. From 12:30 pm to 1:40 pm, I could not reach other ViewQwest customers, so ironically, while power at Equinix SG1 had been normalised, it was ViewQwest’s turn to have internet network problems of their own.
That Equinix SG1 had a power outage is not good at all. It seems Equinix has lots of problems. A significant disruption in 2023 caused quite widespread impact to Singaporeans, as among other things, banking services of DBS were down.
That notwithstanding, it is worse that a network operator could not provide redundancy that could survive a single data centre outage. Their own website didn’t have any redundancy either, something that seemed very trivial to do, and for that not to be done, seems to suggest how little they prioritise the need for robustness and resiliency in their network services.
I switched to ViewQwest in the tail end of 2024, and some friends have asked me about their service, particularly in terms of network reliability, as well as speed/latency to various parts of the Internet. I haven’t had any outages. I did notice erratic network latency, which had great variances, to various Internet sites from the moment I started service with ViewQwest, but this seems to have improved in the last 2 months.
If that question on ViewQwest’s network reliability and performance were to come up again, I am not so sure how I would answer now. It is not so much about the outage this time per se, but that it seems to expose how ViewQwest is incapable of providing basic network service that can survive a single data centre outage.
There was a time, M1 suffered a 3-day outage because something in one data centre broke. I hope ViewQwest wakes up their slumber and take steps to ensure such an outage incident will never happen again, at least not under such circumstances.
View Comment Policy