The recent outage suffered by Death2Spam highlights once again the risks with externally sourced services. Death2Spam is an hosted application that provides spam filtering for enterprise email. They had a catastrophic hardware failure that blew out an entire RAID stack. Putting aside the reason for the actual outage, it’s not hard for the hosted application naysayers to say that it wouldn’t have happened if spam filtering was achieved through an installed application.
Of course the response to that is that hardware failures can happen anywhere – hosted applications just move the locus of failure from within the building to outside. They hopefully also increase the level of professionalism used in both hardware and architecture, thus minimising the risks of failures.
Hopefully Death2Spam has learnt something from the outage and other providers will also look at what they do and start to build in yet further levels of redundancies into their products.
I’m reminded of a debate when I ran a power company. I suggested we move to mobile phones with group voice, if and when they came available (this actually took an extra 5 years to arrive than it should have). The negative argument went – what happens when the cell site goes down? My response was – what happens when our 1-off base station goes down?
There’s no way a small, midscale or even large enterprise’s inhouse service can be more reliable than a utility scale service.
Unfortunately too many hosted services are micro-businesses themselves – worse than inhouse really. Roll on PaaS, driven through utility-scale players!
Excellent comment – and too true!
Cheers Jim, and hope the UK is treating you well!