I started using cloud computing nearly 14 years ago, at the very beginning of cloud computing.
After the initial enthusiasm, I quickly noticed the drawbacks of the Cloud, and it puzzled me why no one was writing about it.
Although I have already wrote about the downsides of cloud computing in one of my previous articles:
Who else could read your private & sensitive data stored in Cloud?
, the order of the main disadvantages has changed.
The main disadvantages of cloud computing that are noticeable to everyone include:
- Data sovereignty concerns
- Highly complex security, especially in hybrid and multi-cloud environments, which often leads to data breaches (this is particularly crucial for EU companies due to GDPR and high penalties)
- Lost of control in problem resolution, especially when the issue resides in the Cloud (you can only open a ticket and wait for someone to address it).
However, the main drawbacks of cloud computing, due to the significant amount of money spent on ‘cloud migration’ campaigns, are very difficult to find.
In this article I’ll name the first three.
1.
100% Vendor lock-in
In many enterprises where I’ve been helping to resolve their issues, I often heard complaints about licensing costs for commercial products, particularly products from major companies like Microsoft, Oracle, IBM etc.
By far the most challenging (and also the most expensive and time-consuming) task was migrating from one database, typically a commercial one, to a non-commercial or at least cheaper, most commonly Postgres.
In the case of migrating to the Cloud, it is inconceivable to perform a migration from one Cloud platform to another, lespecially a migration back to on-premises, as it would require migrating not only the databases but the entire infrastructure (network, storage, services…), where some components are impossible to migrate because there is no equivalent on the target side.
This is true especially for platform-specific services and PaaS components.
2.
Very high cost of Cloud services compared to on-premises
This is certainly something that no one will tell you before migrating to the Cloud.
Cloud marketing typically emphasizes buzzwords like ‘pay as you go’, ‘autoscaling’, savings on IT personnel as the Cloud provider takes care of all infrastructure components, ‘more time left for more intelligent tasks (this part was used in marketing campaigns even over 30 years agoin MS Windows and MS Office productivity suite campaigns) etc.
What is never mentioned is that you will also experience a dramatic increase in costs, depending on the project usually within the range of 20x to 200x.
Describing the main causes of why this happens goes beyond the scope of this article.
Some of them can be inferred by looking at the prices of certain services (and there are thousands in the case of the Cloud, and their prices are constantly changing).
In case you’ve left something running, you’ll have to pay for it, just as if you haven’t written an efficient query towards the database.
Instead of delving into the detailed explanation of the significant cost increase, it’s better to read some of the numerous articles on this topic, such as this Forbes article for example.
The cost increase is so significant that even in the United States, where IT labor costs are the highest, Cloud costs by far exceed the maintenance and development of on-premises infrastructure.
In the case of being persuaded into a multi-cloud approach, managing infrastructure and costs becomes truly a ‘Mission Impossible.’
A good way to obtain accurate and unbiased information is through indirect means.
In the following article, for instance, the CEO of NetApp (one of the top 5 enterprise storage vendors) explains in the context of business results how customers are extremely price-sensitive, in this case, by choosing budget hybrid-flash instead of flash array storage in the Cloud.
If you talk to IT engineers, you will find that, in most cases, their first choice is on-premises, where they have full control over everything at significantly lower costs.
Here is a link to the team at Sofascore, which has millions of online users (the text is in Croatian, but you can easily translate it).
https://www.debug.hr/sofascore-sto-stoji-u-pozadini-aplikacije-s-milijunima-aktivnih-korisnika/
In the documentary film “The World According to Amazon” one part that stuck in my memory is about Amazon’s revenue and profitability.
Out of every 10 apples of revenue, 9 come from the Amazon webshop, and only 1 from AWS.
When it comes to profit, out of every 10 apples of profit, 9 come from AWS, and only 1 from the Amazon webshop.
I think that says it all.
3.
Performance / Cloud slowness
Cloud is slow, and it’s not created for high performance from the beginnning.
Latency is just one of the reasons for slow Cloud performance.
There is an excellent article from Informationwek which starts with: „Latency is the big killer of cloud computing, especially public cloud computing”.
You can find the entire article at the following link:
https://www.informationweek.com/it-infrastructure/edge-computing-eats-the-cloud-
or in the following article related to Multicloud deployments:
https://www.infoworld.com/article/3690333/why-is-your-multicloud-so-slow.html
The best way to verify the slowness of Cloud computing is to conduct tests in both on-premises and Cloud environments yourself.
Since conducting such tests is often not possible, the best way is to obtain facts indirectly.
Although Larry Ellison’s speech during the Q3 Oracle’s quarterly earnings call may be considered biased, still reveals the main drawbacks of Cloud computing:
- network latency/slowness
- virtualized CPU slowness
- virtualized storage slowness
You can find more details at the following link:
https://www.crn.com/news/cloud/oracle-q3-earnings-5-things-to-know?itc=refresh
The best way you can learn about comparison of Cloud vs. on-premise computing speed is from benchmark tests conducted by various companies.
In this case, it’s a benchmark of ClickHouse vs. Amazon RedShift.
As part of the benchmark tests, a test was conducted for ClickHouse on-premises vs. ClickHouse on AWS (the section that starts with the title: ‘Amazon Hardware Benchmark’).
Although this is a test of only one component (in this case, a database), it’s evident that on-premises computing is, on average, 3 times faster than AWS.
You can find the entire article at the link below:
https://altinity.com/blog/2017-6-20-clickhouse-vs-redshift
If a realistic test with a larger number of components (databases, middleware, etc.) were conducted, the result in favor of on-premises would be on average 10x to 1000x faster compared to the Cloud.
Summary:
In today’s world, especially after the COVID crisis and the wars that have shaken humanity, along with disruptions in the oil market, transportation, logistics, and the like, old rules of globalization like ‘produce where labor is cheapest’ or Material Resource Planning (MRP) rules for establishing just-in-time supply chains are no longer applicable.
Translated to IT industry, data sovereignty, self-sufficiency, and complete control over all vital parts of the IT system provide companies with resilience that is most evident during a crisis, as you rely primarily on yourself.
That’s precisely the purpose of this article – to highlight some of the disadvantages of Cloud computing alongside its advantages, which IT portals and various marketing campaigns have failed to mention.
Important update:
Shortly after I finished the article, breaking news emerged that Amazon had earned one billion US dollars through a secret price-raising algorithm.
The article was published by Reuters and Forbes, which are considered reliable sources of information.
If this is true, and assuming we live in an ideal world, such companies would be considered criminal organizations, while in reality, they are called tech leaders.
You can find out more on the following links:
and
Comments