Everything as a service
Venkat Swaminathan
Ideator, Innovator, Hacker: Connected solutions, AI, IoT, Home Automation, Prototyping
I drove with my daughter yesterday to a small town in the path of totality of the solar eclipse so I could get some good photographs and have the experience. It took us about 2 hours to get there. We had a lot of time to find parking, settle down under a nice tree, set up the cameras, eat our packed lunch and generally chat with others. The return trip was sheer misery. It took us nearly 5 hours to get back. The roads were not planned to handle that kind of traffic. Which got me thinking about how everything-as-a-service (EaaS) is gonna work in the future when we rent everything and own nothing.
Things that did not work well yesterday for me in the face of the unprecedented combination of events:
- Cell coverage: everyone had 3 or 4 bars of coverage but no one could connect to data services. There was simply not enough capacity for the network to handle 25,000 people in a town that rarely ever sees more than 2000.
- Google Maps: between the lack of data coverage over a large region and the huge crowd, Google Maps kept switching ETAs (it would show 1.5 hours to destination and then switch immediately to 4 hours, before it changed to 2 hours, etc.) We were driving at the speed limit on a road that Google Maps showed as red. We were sitting in traffic for 20 minutes on a road that Google Maps showed as blue. I wonder if someone at Google is reviewing the events from yesterday -- I think it'd be very interesting.
One prediction has that car ownership will decline as more people use services like Uber and Lyft today, and driverless cars tomorrow that you can summon from an app. For most of the days that would work fine. What happens when you want to view a rare event like a solar eclipse?
If past experience is any indication, then almost no system today, in any domain, can handle a theoretical peak load that is feasible, even if highly unlikely. Banks cannot provide cash on demand to EVERY customer. The power stations (in most countries/regions) cannot provide demand to EVERY customer using EVERY appliance at the same time. Grocery stores cannot stock enough milk and bread for every customer before a predicted natural event like a storm. There are enough examples. The way we deal with that is through alternatives and planning. We plan both load and capacity to match availability. If I can predict a run on a bank, I would stock up on cash (or gold?) in advance by withdrawing a little a day. You can read more about the demonetization event in India in 2016 as a recent example of how people underwent some serious hardship.
If you are a consumer of services then here's a scenario that I visualized for an event like the solar eclipse if no one owned vehicles. There'd be those who had planned the travel and had signed up for the ride service in advance. There'd be several though who would have made plans at the last minute or for some reason not be able to get a car. All those autonomous cars on the road were not going to be able to get anywhere. There'd be huge delays, irate customers waiting for their car to show up, accidents. What would be your perception of transportation as a service after such an event? Would you understand why you didn't get a car for 3 hours? Or would you switch providers? Would you own a car for special situations like these?
If you are a provider of services then what's your plan to respond to a "perfect storm" like the event of yesterday? Will your service delivery mechanism scale to handle the load? Do you have enough resource capacity to respond to the demand? Are you going to throttle your services and/or rely on your circuit breakers? What about the intermediaries you use to deliver your services to your end users (e.g. if you Acme Rideshare, does your cloud infrastructure platform provider have enough resources)? So many questions.
I will look to posting an update to this post in April 2024 from Carbondale, IL. Google for it if the context is not apparent.
(all opinions in this post are mine and mine alone and do not reflect those of my employer or any other organization)
Sales Architect at Informatica for Data Management Cloud Solutions
7 年Very nice perspective - Venkat!
Very well said Venkat, we have slowing walked away (Carrier Infrastructure) from .999999 reliability -dedicated network to wider IP protocol based service delivery (It is OK for IP calls to get dropped once in a while, Good is OK from Six-Sigma). This is a great observation, how will we deal with Extreme events. BTW interestingly WhatsAPP worked during Chennai Flood not voice network. Thanks for sharing, gives me few ideas to evaluate,