Stormy Weather

It seems that the rush towards a Cloud-based future I mentioned last week has had some crowd control applied to it recently with a series of unfortunate events (including Microsoft this week) that have reduced the need to tear gas the angry hordes clamouring to climb about the cloud carousel.

 The big boys are spending billions on secretly building larger and larger climate change accelerators, ironically called cloud platforms. The Final Programme to inevitable move everything to the Cloud has unstoppable momentum this year (a bit like COVID-19), but a few drops of reality coughed in our faces are causing many of us to check our PPE.

 All the cloud providers have had brownouts (like blackouts but smellier) of their online services, and will continue to do so. Expecting a cloud service to have 100% SLAs is not only naive, but also impractical. Check your XaaS service contracts to see how poor they really are.

 This won’t come as a surprise to those old cynics like me who have learnt to steer clear of bleeding edge technology solutions for anything serious until at least three major disasters have happened. Preferably to other people. So does this mean it’s safe now? No one who has been in IT for any length of time would use the words ‘computing’ and ‘safe’ in a sentence without a ‘not’ between them. You can build reliable and logically flawless systems but, unfortunately at some stage people get involved and it all goes Pete Tong. Ask the Architect in the Matrix.

 With Cloud this is not only true, but you also have a long piece of string between you and the remote server farms that are more vulnerable than is commonly assumed. Workmen the world over are eagerly looking to cut through the wrong cable with a well-placed shovel, and the Axis of Evil are looking to control the data flows with technology they’ve nicked off the Axis of Incompetence. The ISPs, struggling to cater for the exponential traffic growth are gleefully throttling your pipes under ludicrous ‘fair play’ small print, while leaving too many SPOFs (Single Points of Failure).

 So, my advice is to treat Cloud like any other IT service:

  • Halve the claimed benefits
  • Double the proposed costs
  • Add six months to the delivery time
  • Make sure plans B & C are in place
  • Wash your hands before and afterwards

John “Hullo clouds, hullo sky” Moe

Leave a comment