What is Cloud Computing
Bibhash B.
CTO @ Mindshow | Machine Learning | Deep Learning | AWS | Harvard Business School Online
A brief history of the cloud and why it’s vital to you.
Cloud will be synonymous with the word Web. Can you think of a business today that’s not on the internet? If you don’t get on the cloud your company could be falling behind quickly.
This article will help you to get what the cloud is even if you don’t know much about technology.
What is the Cloud?
To understand the cloud, let’s talk about the computer in your home. It is primarily composed of 3 things. A storage system for storing data. Some chips on the motherboards to perform tasks (tasks like moving data around or running some algorithms), and a user interface device to show the data to the user (computer monitor) through which the user can interact with the system (keyboard and mouse)
The first thing that people started needing more is storage capacity. Back then, like 7 years back, you probably had more external hard drives than you had computers in your home. Cameras started to increase in megapixel count, and all those home videos of birthdays and family trips on digital video cameras needed a lot of space to store.
Managing hard drives and worrying about backups in case they failed were a constant headache. If you ever bought Seagate hard drives, you already know.
Then Dropbox came along where you could buy some storage space online and upload your files to Dropbox. They maintained backups so you didn’t need to worry about losing your data or keeping manual backups. They appeared as another drive on your computer just like if you had plugged your external hard drive in.
No more external drives to manage or USB cables to wrangle or running external power supplies. Your data on someone’s else servers under your name.
That was storing data on the cloud
The one thing you worried about was security and that has come a long way. We’ll talk about that later in this article.
Today we pay $1/month for 1TB of data to Google to buy storage on Google drive. A 1TB Western Digital hard drive would cost about $55. That’s buying 4 years worth of storage on the cloud for which you don’t need to worry about backups, hard drive crashes, or worst — the problem of whatever version of windows would be there 4 years from now not recognizing your USB drive anymore!
While consumers started offloading their storage to the cloud, companies started thinking about how to offload more than just their data.
A story of servers and data centers
Big companies have massive data centers — buildings full of servers. Medium-sized companies have 1 room full of servers called the server room. In addition to data, companies have another problem — they need to estimate how much compute power they’d need on their servers. This is the most challenging part. Adding more hard drives to satisfy storage needs is the easy part.
Servers in a company serve the internal world — the employees, and the external world — the customers or users of the company’s product.
When the company is being set up or before a product is launched, someone (usually the CTO or the VP of Engineering or the CIO) would need to estimate the number of servers they’d need to be able to handle the estimated number of customers.
It is a big deal to order servers. They are huge, they are expensive, they take months to get delivered, they need to be installed by experts, their physical backups need to be installed at the same time at some other location, and the proper cooling for servers need to be installed in the room (Servers run 24/7 and they need to stay cool…very cool. This is why you can see your breath in a server room.) along with thousands of wires snaking through the walls that are feeding-tubes for all the computers that these servers are serving inside the company.
Add to that, the cost of buying expensive server software licenses, paying exorbitant fees for database management systems, electrical panels to support the high wattage, extra security system for server rooms and hiring the right people to set up and maintain these servers, networking, firewall, and work the software.
In essence, this is the company’s IT department. When you buy servers for your company you also have to plan for adding a department!
The whole process of provisioning servers on-premise took 2–3 months. You just don’t change your mind in the middle of this process. You needed to be able to estimate well. After going through all of this you wouldn’t want to be wishing for one more server! Of course, people usually erred on the side of more than less.
Generous estimates led to a waste of money, a large number of servers mostly sat idle while the user base grew to the estimated number.
On the other end of the spectrum- a lack of upfront capital leads some companies to order less than what was needed. In which case, even after spending all this money they weren’t able to satisfy all of their customers during peak time.
Some company’s products had seasonal demands. Maybe their product demand peaked during the holidays. This meant that users would grow during those months drastically. The company had to provision the servers based on these events otherwise users would get server busy signals or server unavailable error or request time out errors when they needed the product the most. Frustrating your clients at any time, especially during peak season is death for a business.
All those servers sat mostly idle for the rest of the year and the company just kept paying for the upkeep, maintenance, patching (software updates), physical cooling and the rent for space to house them.
This is like buying a bus and driving it year-round because you want to be able to take your family and your extended family around town when they visit you over the holidays. All other days of the year, you’d drive that bus to the grocery store all by yourself just to pick up a loaf of bread. Ridiculous but that’s what hosting servers and data center on-premise can look like for small businesses.
Many small companies don’t have the money. So they’d leave it to third-party hosting services to manage their servers and data centers. They get roped into expensive plans. Moreover, they’d need to physically call or email someone when things need to change. With most of these third-party hosts, scaling on demand is not immediate. You have to call it in and go through the process of a subscription plan change.
For large businesses, the problems are in maintaining the servers and providing fault tolerance to guarantee availability round the clock and building out data centers across the globe to be able to serve customers quickly in different parts of the world. In most cases that’s not even their core business. Their core business maybe is to sell diapers online!
Imagine investing in poultry farming and hire people who can maintain the farm every day so that you can make an omelet for breakfast!
Enter the Cloud
How would you solve these problems for businesses? You buy up warehouses in Phoenix, Nevada, N. Virginia, Ohio, and all the places where real estate is cheap, and then install a bunch of servers on all these locations. Run high throughout fiber optics cable to connect everything, secure the facility, secure the data at rest and the data on the move. Then start renting them out to businesses. You manage the headache of buying, installing and maintaining these servers while the businesses can focus on their business!
That’s what Amazon did in 2005. Besides selling books online, they spun a new business venture under the Amazon banner and called it Amazon Web Services (AWS)
Now, small businesses could start to make their omelets without ponying up millions of dollars for a data center just to test out a product idea. The entry barrier to software business was lowered, the viability of the business increased, and the time to market which is a critical factor for small businesses was cut down drastically!
Moreover, you only paid for the time you used the servers. If your app was love-luck roulette and people only used it on Valentine’s day then you didn’t need to pay for servers for the other 364 days! (This is not a joke, there are many apps like this…my friends have developed one and they made 17K on their first year from the iOS app store from people using it primarily on Valentine’s day!)
In addition, AWS would scale the servers. If you have a few users on one server and you suddenly see a surge to a million users, AWS would provide the extra servers to handle the demand right away without a glitch! When the demand died down, AWS would release the servers for someone else to use while downscaling you to just the number of servers you need.
In 2008, Google came up with its cloud services called Google Cloud Platform (GCP), and in 2010, Microsoft released it’s cloud services and called it Azure.
There have been many smaller cloud providers but the 3 main ones today are AWS, GCP, and Azure. According to the technology analyst firm, Canalys, AWS has the largest market share at around 80% while Azure is at 15% and Google at 5%.
They all pretty much do the same thing. AWS has a 4 year lead on these guys and it shows. AWS is vast. They have about 200 cloud service offerings and it is growing every day.
Google is more narrowly focused in its cloud offering. They don’t offer as many services as AWS but they have an edge in that they own Tensor Flow which is a framework used by most machine learning programs.
Microsoft has an edge in that it is the largest PC market. They are leveraging that edge to make it easier for their existing enterprise customers to seamlessly integrate with Azure.
The race for Cloud domination is well underway and AWS is leading the charge. Google and Microsoft don’t break out their Cloud earnings separately in their quarterly earnings report so the exact growth numbers are still a little bit of a mystery from both of these companies. But, Microsoft’s Azure is the fastest growing in the Cloud business today.
Enterprise customers have so much data and infrastructure that it’s not as easy as to flip a switch and move to the cloud. Just the sheer amount of data that needs to be transferred by an enterprise client can take years to upload to the cloud even with the fastest Internet provider today.
AWS introduced a special truck called a Snowmobile to solve this problem. It’s a huge truck with huge hard drives. They’d send a highly secure truck out to your location so you can transfer the data on the hardware that then drives out to Amazon facilities. The data is then uploaded to AWS using Amazon’s backbone infrastructure which is much faster than what’s available to companies over their LAN line. The process can still take a few weeks but, its better than a few years!
One of the biggest challenges for moving a company’s whole operation to the cloud is the amount of investment that has already been made to the on-prem infrastructure! For small businesses who never had the money to invest in an awesome secure on-prem data center, the cloud is immediately attractive but for enterprises who have spent a few hundred million dollars setting up and maintaining their own data centers, its a difficult decision.
There are concerns about security and compliance, which we’ll talk about in a minute. They’d need to start paying millions of dollars to AWS/GCP/Azure just to support their global customer base. What are they going to do with the millions of dollars of infrastructure that they already have?
Besides petabytes of data transfer/storage and expensive infrastructure, there are other sensitive factors to consider — data privacy regulations (e.g. HIPAA) and country-specific laws where data needs to reside within certain geographic boundaries.
All these issues are being worked on. They are not deal breakers to the cloud revolution. They’ll get sorted out in time. AWS, GCP, and Azure are all HIPAA compliant today which means the healthcare industry with all its data about you and your health are eventually going to be moving to the cloud.
Let’s talk about Security
Cloud security and data encryption have come a long way. Cloud has had its share of data breaches in the past. Today, the security landscape on the cloud has evolved. Data on the cloud is oftentimes more secure than data on-premise.
Enforcing security on-premise comes down to the security personnel you’ve hired to keep track of security. In a company, this person is the IT dude. This dude has plenty of other headaches.
The primary reason that most companies are vulnerable today is that they fail to keep up with the latest software patches on their servers which makes them sitting ducks.
On Cloud, the security scales automatically with your data. On-prem, if you were to add extra resources to handle last-minute loads there is a chance that in order to get it up and running quickly some security practices can get compromised or left for later after the fire is out.
Having said that, Security is not a turnkey solution. Security is still a shared responsibility between you and the cloud provider.
For example, AWS provides Multi-factor authentication (MFA) but it is up to the user to turn it on.
You still have to do your part. But rest assured that the physical infrastructure, firewall, software updates, intrusion detection across all of the hardware and managed services come from the cloud provider.
Like security, compliance is also a shared responsibility between the Cloud provider and you. But, having a partner like Amazon, Google or Microsoft while navigating these waters can save a lot of time and money.
Google claims that its cloud security was designed by a 700 member engineering team. Here is a white paper on Google’s cloud security
Conclusion
If you’re a small operation, the cloud will give you a virtual data center and an IT department wherever you are. This means you can open your office anywhere, and your datacenter comes with you!
For bigger businesses, the cloud will save you money by cutting costs of setting up server rooms, data centers and hiring an IT department. It gives you the reliability, throughput, redundancy, and availability as part of the contract. It makes it easy to share resources between your employees. It allows you to hire off-site personnel and integrate them into your office space easily. The virtual ‘IT department’ on the cloud meets everyone where they are.
Cloud has progressed in leaps and bounds in the last decade. Cloud-first is a revolution that is sweeping across the globe. This is probably the best time to start thinking about what are you going to do with the cloud, have a plan and start the shift if you haven't already.