An SLA Detente?
January 5, 2012 Leave a comment
The latest ‘cloud computing’ newsletter pointed at Alistair Croll’s Top 12 Cloud Trends Of 2012 which makes an interesting read; particularly interesting, I found, was his ‘Trend No 8: An SLA Detente’ (I had to look Datente up, turns out it means relaxation…).
In his article Alistair suggests, and I paraphrase, that customers’ expectations from the SLAs provided by cloud vendors are unrealistic and he suggest the point that Car makers don’t provide insurance as part of the deal to purchase the car, but rather customers buy insurances they deem necessary separately.
This had got me thinking – the man does have a point. to a point.
I would suggest that analogy stretches further, and actually taking about the warranty is a better viewpoint – Consider you’re a logistics business in need of a fleet – you would make your research and settle on a car from a maker with a good track record of reliability and service.
You might pay premium for these, but you asses the risk and suggest that buying a truck from a reliable manufacturer is worth more than buying one from a less robust maker; I don’t think anybody believes it is possible to buy a truck which ‘5 nines availability, and even businesses that relay on these accept a certain amount of ‘down time’; of course we expect a good maker to have a good service network, that will see our truck fixed in the shortest amount of time possible (and at the first attempt), and we would almost certainly expect a courtesy car/van/truck whilst ours is in the garage, but if it takes half a day to sort this out, in the main, we accept the fact.
Now – I don’t suggest that cloud platform are equivalent to cars and certainly when thinking of the Windows Azure platform I would think closer to the space shuttle – with tons of redundancy and a lot of ‘big brains’ behind it, and so – the chances for failures are indeed much smaller than those of a car or a truck, and recovery is much faster, and certainly at large IT operations in general we tend to aim for the mythical 100% availability and try to hit that 5 nines promise, largely because we believe we can, but I do think I agree with Alistair that organisations increasingly feel more than comfortable with the SLAs on offer, accepting the cost/benefit analysis behind it and – more importantly – accepting that in most cases they could not have achieved better or even similar in most cases!)
I also agree, and would like to emphasise Alistair’s last point on this topic which is that cloud vendors, and certainly the Windows Azure platform, provide a lot of capabilities that allow solutions respond to any issues that emerge and so building high availability solutions is actually much easier on Windows Azure than it is on-premises, or – as he had put it ”In 2012, we’ll realize that the providers have been trying to tell us something: You can have any SLA you want, as long as you code it yourself and find a way to turn risk into economic value.”