Baxteen
Active Member
- Joined
- Sep 27, 2020
- Messages
- 776
the outages shown are user reported outages rather than system reported outages. so chances are local outages for App/internet banking/ATM/POSIt would depend on what we class an outage as.
In my current line of work, an outage is defined as the bank's system not being available.
Everything else is highly localized as in the ATM or EFTPOS card machine cannot get network connectivity, Eskom, etc.
I presume those outages shown are as in- the bank is totally offline, apps don't work, people can't draw cash at the ATM, etc...
ABSA, followed by FNB and Capitec are notorious for it, especially around payday and especially around January payday.
In my industry, the first quarter of the year is usually when shit hits the fan...
As for what these outages are caused by:
* RAID failures (all the SSDs failing at the same time or closely after each other)
* Hosting centre going down (yes looking at YOU MTN Business and Internet Solutions)
and less commonly...
* ECC RAM failure combined with server mainboard failures
* Network equipment failures (this one is a far more common occurrence in our office which does not affect the banks)
hardware failures and hosting failures are less likely causes as even the smallest banks have a live failover site with a different host as well as a third Disaster recovery site hosted with the primary host in a different location. so in the cases where I worked it was Primary host in teraco back up host MTN and Third failover also being hosted by Teraco but at a different site.
hardware failure that I have seen impact up time was a switch that died at Vodacom that prevented traffic being routed correctly so that caused failures as the routing was messed up and lead to response times being too long and the machines showing a failure.
Firewall fuckery is more likely to be a cause of failure than hardware related failures.
what I have also seen in my time working in the industry is that the online banking, app banking (so API server) inbound instant EFT/SWIFT transaction, ATM/POS transaction and then inbound and outbound file based transactions (EFT/ DEbit order) are all hosted on separate hardware instances. so if ATM is down because the mainboard failed, app banking is not impacted for the few seconds it takes to switch to backup/DR
but that was just where I was, and that was a small opperation that handled the kind of banks most of us here forget about. obviously cant disclose what the clients were but trust me they are all small with very small user numbers. there was one bank that has a large user base that is rapidly growing and will be seen fighting with the "big" banks soon. as for the others there are at least 3 that no one here has ever heard of as they have a specialised user base thanks to religious restrictions or a spesified use case. no one here are Farmers or TAxi bosses so strong doubt those banks are even known about.