Edge Compute is lot more than simply reducing latency for applications.
I get asked very often, what do you do for a living? I respond by saying I work for a Data center Provider. The follow up question is often what exactly does a Datacenter provider do? I respond by explaining that Data center provides datacenter space to their customers. People associated with IT profession get it, but non-IT people still have a blank look on their face. I explain further to non-IT folks that Datacenter providers lease datacenter space to customers who don’t have the expertise of building and running a datacenter, or want to focus on their core competencies and leave it to Datacenter providers to manage their day to day datacenter operations. It’s very similar to an individual deciding to rent a house, rather than buy one. However, the most important factor is clients can leverage any of the various DCs that a Datacenter provider owns to bring their applications close to their end users. It would be a huge undertaking for an organization to build 40 or 50 DCs just for this purpose. The end purpose is always to make the end user experience better. End-user experience consists of applications latency, applications response, and uptime. We often hear the word “Edge Compute” which of course reduces latency, by bringing the applications close to the end-user, but it does lot more than that. Let’s dig into it.
Three years back, every major corporation’s strategy was to make edge compute nodes close to their scattered end-users. We saw this with AWS, trying to reduce end user latency with buying all CO (Central Offices) from Telco providers and turning those into edge nodes. Remember those ESS switches in Telco COs in the 90’s that used to take up the whole room? Well, over time those switches became so small, that you could put several of them in a 42U rack. Telco providers decided to use that extra space in the COs to turn them into small scattered datacenters to be profitable. Major Enterprises started grabbing those facilities with power, cooling, and generators to turn them into edge computing nodes. The concept was to make their end-user experience better on e-commerce sites, by moving compute closer to their end-users. It just made the application response much faster for their end-users. It is well known that users wait no longer two to three seconds for a website to load, before they head-off to a competitor[i].
How will Edge compute help me sell an extra gallon of milk?
While working for one of my former employers, I had the opportunity to talk to some major convenience store’s technical staff, about their application latency requirements, and leverage edge compute for their applications. Application latency means different thing to different people. For self-driving cars, Stock Exchange order executions, Artificial Intelligence, and/or Machine Learning each ms. can matter. During the discovery session, it came up for a convenience store chain processing a transaction in 30-40ms vs. processing it in 5-10ms has practically no impact.
If you think about shopping at a grocery store or a convenience store, would you, as a consumer, care that it took their credit card approval transaction an extra 30-40ms (that’s 0.03 to 0.04 secs)? Most probably not! So, what’s the benefit to enterprises that are not concerned with low latency, but can benefit from edge compute or Network Hub strategy? How can Network Hub or Edge compute help a business where a little latency doesn’t effect their clients? How can Network Hub or Edge Compute help a convenience or a grocery store sell an extra gallon of milk?
If you ever walk into a restaurant or one of the convenience stores, what is hard to miss is that in the back of the store, close to rest rooms, you will see a server lying in plain sight. These servers are usually doing the credit card processing, inventory management and other critical back-end office functions. This mission critical infrastructure is lying in open, suspectable to all the environmental hazards around it. An unsuspecting patron can drop a Large Soda on it, by mistake, or it can gather grease from fryer in a restaurant. Any of these incidents can take down the mission critical applications that the store or restaurant depends on to keep their business running.
In scenarios like these, imagine you can move infrastructure to one of the datacenters nearby (edge compute), where it is in controlled environments, away from the hazards of grease and/or soda spills. Now, you are reducing the risk of business being out for several days while the server is swapped, operating system reloaded, and applications recovered or installed from scratch. What about all the potential data loss and business opportunity lost because the convenience store or restaurant was unable to process credit card transactions or automatically re-order inventory that was diminished during the outage of equipment?
Imagine using Edge compute in the scenario above. You can take several convenience stores or restaurant locations in the local area and virtualize their infrastructure on one physical server. Not only will you protect the server infrastructure in a secured environment, but also be able to be environmentally conscious by reducing your carbon footprint! One server can host several local stores’ applications, and send the data over dedicated links into local stores. The store or the restaurant would simply have a router mounted on a wall away from all the environmental hazards. Wouldn’t that help optimize a business and reduce the potential of catastrophic event like a server going out for several days?
领英推荐
I remember one day I was presenting on my computer to a customer, while working from home. This is long before COVID hit, and it became a norm for a lot of us. While I was presenting, cleaning crew came in. My networking equipment was in the basement with a 12 port power strip plugged into the power outlet. The power strip had many ports available, but somehow the cleaning crew, despite all the empty power ports decided to use “ Yes, you guessed it RIGHT”, the plug where all my equipment was plugged in.?Of course this kicked me off my customer meeting, as my whole network had gone down, and after plugging my network back in, it took several minutes for me to login back into my meeting, apologized profusely and continued my presentation. So, yes things like this happen, and when critical business equipment is kept out of plain sight, in a controlled environment, it can help make it more efficient and less prone to environmental factors surrounding it.
I am sure you have heard the joke about the hospital where a patient died every Sunday at noon, at a specific bed. The joke goes something like this :
In a hospital's Intensive Care Unit, patients always died in the same bed, on Sunday morning, at about 11:00 am, regardless of their medical condition.
This puzzled the doctors and some even thought it had something to do with the super natural powers at play. No one could solve the mystery as to why the deaths occurred around 11:00 AM Sunday, A worldwide team of experts was assembled to investigate the cause of the incidents. The next Sunday morning, a few minutes before 11:00 AM all of the doctors and nurses nervously waited outside the ward to see for themselves what the terrible phenomenon was all about.
Some were holding wooden crosses, prayer books, and other holy objects to ward off the evil spirits. Just when the clock struck 11:00, Pookie Johnson , the part-time Sunday sweeper, entered the ward and unplugged the life support system so he could use the vacuum cleaner.
Joke aside, the point I am trying to make is edge compute is responsible for reducing latency and lot more! It can not only reduce application latency for latency sensitive applications, it can also help reduce the environmental factors to reduce the business downtime. Sometimes keeping the infrastructure up and running is literally and figuratively the bread and butter of a business.