“Cloud bursting,” streamlining, & preparing for huge demand

“Cloud bursting” is a new term that describes leveraging compute resources external to an organization, when the demand is greater than the organization has capacity for.  Basically it is a process that provides additional resources that an organization doesn’t typically pay full price for the additional resources initially, but pay based on a consumption model when those additional resources are used.

Is IT the only organizational unit that does this?
Nope, not by far.  Scaling quickly during a time of increased demand isn’t really isolated to just infrastructure and technology units.  It has been around for a long time.  In fact this practice has been around a long time, longer than most IT departments have been around.

Holiday Shopping (Seasonal Workers)
I started working at a department store my senior year in high school.  I know, it wasn’t what I wanted to do, but I was a kid that liked having money, and I worked.

As anyone who has ever been to the mall during the Christmas season, there is a lot of traffic.  I mean a LOT of traffic.  During the regular part of the year there when I went to work, it was really anyone’s best guess what the day would be like.  Were we having a sale? Was it going to be a slow night? No way to tell.  As a result, we staffed our department accordingly.  We had enough people to cover, while keeping our expenses down.  It really didn’t make sense to have 5 sales people working, when we would only get 6-8 customers over a 5 hour period.  Because I worked in electronics, it wasn’t really a “let me ring you up” type job, but more about learning what the customer needed, and then help them with the right products for their needs.  We would typically have 1 or 2 employees working, depending on the night/sales/etc, to make sure we had enough coverage.

Every year around early October, we would start to interview people for “seasonal positions.”  These workers were hired to be able to help with the onslaught of additional traffic/customers that would always happen during November & December.  This was a carefully planned process that had been successfully implemented year after year (at least by the time I was introduced to the process).  The retail industry seemed to have a good handle on when to hire, how many to hire, and so forth.  I’m not going to say that we always had the coverage we needed.  I’m also not going to say that those of us that worked didn’t work harder during those times.  I am going to say, for the most part, the number of people hired, in conjunction with the coverage provided, it seemed like a happy medium.  I will admit that not every customer was the happiest when they had wait to be helped, but overall, sizing seemed to be appropriate in the store I worked at.

We also moved to accepting credit cards to help with the purchasing process.  We gave customers the ability to call in orders (late 80’s folks/early 90’s) and simply pick them up without ever having to talk to a sales associate.  We had our own process to burst, allowing us to keep up with the enhanced demand during the holiday season.

Utility Companies (Disaster/Contingency Workers)
So winter weather has knocked out electricity in the Northeast.  I’m pretty sure that many of the utility companies don’t have enough employees to accommodate for the outages.  I know that when Hurricane Katrina hit the Gulf Coast, that the utility companies didn’t have nearly enough employees necessary to bring everyone’s power back online immediately.

I grew up in a house with a father that was a career utility worker.  There was always work to do, even without a disaster like a hurricane or a massive winter storm.  From time to time, there would be terrible weather, and overtime hours would have to be worked.  Other times there would be some major disaster somewhere, and he would go to help.  From what I understand about the “helping process,” many utility companies had agreements among themselves to help each other out in times of need.  Having not been a member/employee of that industry, I can only speculate about the relationships between these companies.

I remember when Katrina hit, there was a massive amount of electrical workers that descended upon New Orleans in an attempt to get service back to normal.  I’m not really talking about anything from a government perspective, but from the local electrical company.  It was really amazing to see how that company orchestrated getting workers to the damaged sites, and coordinated getting electricity back in service.

One of the major differences between utility companies during a disaster, and retailers during the holiday season, is that there is no way to plan for the increase in demand, at least not with 100% accuracy, and as quickly.  Mother Nature typically gives us enough information to know when weather is going to get bad, and when to expect certain types of weather are traditionally are going to occur.  There isn’t really any way to know ahead of time how much damage is going to occur.  Also the “ramp up” time, especially when something like a hurricane or a blizzard, is often very short.  Sometimes only a few days.

Another difference, is that retailers have a huge amount of control when it comes to the layout, flow of traffic, amount of stock, technology used, and so on.  Retailers spend a lot of money on how to lay out their stores/departments to make sure of maximum efficiency.  They have the luxury of being able to change things after a bad season to make the next season better.  Utilities on the other hand don’t have the luxury of changing their “playing field” as much.  There is too much cost involved.  Wouldn’t it be great if everyone had electrical power lines buried underground? It would, but it would be very expensive to change so many existing neighborhoods.  Utilities have traditionally moved power lines underground in new neighborhoods, in an attempt to mitigate outages from downed lines.

I’m not saying that utility companies don’t plan accordingly.  What I am saying, is that relationships are put in place to take care of things afterward.  I can only estimate how much more utilities would cost if more people are employed “just in case” of an emergency.  In 2011, the Gulf Coast was pretty much spared from a bad hurricane season.  How much more would costs have been if the utility workforce was twice what it is now, when there wasn’t really as much work to do, resulting from a disaster, this year.

How do these compare to Cloud Computing?
Depending on what type of workload that is being run, cloud bursting can fit into these models, or many more.  The big difference between the two examples I used, is that one had the ability to streamline their processes, and the other didn’t (at no fault of their own).  How do these directly relate to organizations today?

How many organizations out there would like to employ cloud bursting, but have legacy applications that don’t scale well?  Legacy applications, like power lines that are too cost prohibited to move, can be a huge hindrance to scaling.  In a situation where a legacy application has to scale, it could take a while to scale it to the cloud.  This is just like having to bring a bunch of utility workers in from out of state to help get electrical service restored.

In the seasonal worker model, they not only scale by hiring temporary workers, but have the flexibility to change the way they do business.  Whether they employ techniques for proper customer flow, product purchasing, or implement a buy online with store pickup option, retailers have used methods to help customers more efficiently.  The next time you go into your favorite retailer’s store, take a look at the layout.  Count the number of employees are working during non-peak and peak times. Also check out their web site, see if they have a buy online/store pickup offering.

The important thing to remember, to effectively use cloud bursting, make sure the services that you offer can easily scale to the cloud.  Sometimes you have to change the way you offer services to be able to handle the demand most effectively. In some cases this can be cost prohibitive, depending on the amount of investment in legacy technologies.  Sometimes a fresh start is costly up front, but more cost effective over the life of a business. Cloud bursting can be a very effective tool to scale, just make sure that your services will scale well to the cloud if you want the expectations to be the same as they would if they were offered from your private cloud.