5 ways your database can kill your business?

5 ways your Business can be crushed by the competition

And discover how you can prevent this.
The stories depicted here are all real, I simply changed a few names and removed unnecessary content...

#1 In a race, the slowest stays last, or dies...

When Jerome, one of my clients, called me this had been his daily routine for a week:

Reboots
Crashes
Downtime
Slow queries leading to horrific user experience.

They were hardly reaching 47,191 orders a day and the server kept freezing and crashing.

Customers were leaving their carts, leaving the site and making their orders somewhere else instead.

They were fed up and frustrated.

So they went shopping somewhere else with a better and faster user experience. Spending their money with a competitor that was more than happy to indulge them.

And those customers were so angry that they might never come back.

In fact, they even sent complaints on how bad the service was on a daily basis. The customer service was receiving hundreds of emails everyday.

And this is when Jerome finally decided to call me... after ignoring all the warnings.

It took 10 days. There was doubt and fear but in the end…

He went from hardly 47,191 orders a day and a crashing server…

To an easy 66,567 orders and 70.89% increase in revenue…

It took a lot of hard work. But that was mostly on my part.

Now, let me ask you a question.

Do you think you stand a chance to win a race when you are constantly at the end of the long list of contenders?

Obviously, no.

Your app is fast today. But do you think about tomorrow?

I've seen it time and time again.

The project launches. A few weeks in, and the first signs of slowing down appear. And most of the time all my clients ignored this first warning.

But it's usually the last.

What do you think your competitors are doing?

Lets be honest and realistic.

Your competitors hire the best experts they can find to keep their business running.

When processing payment, Nicolas wanted the process to take no longer than 300ms.

Jerome wanted less than 100ms.

Maybe big names think differently:

53% of the time, visitors to mobile sites leave a page that takes more than three seconds to load.

(Think With Google)

A one-second delay in page response results in a 7% reduction in conversions.

(Neil Patel)

Do you really think you can afford to keep your app slow?

But it's not just about payments. It's the whole experience.

Antoine was desperate: paying customers could not enjoy new features he was developing for his business because it was too slow!

Let's see what big names

have to say about usability:

UX stats show that 88% of online shoppers say they wouldn’t return to a website after having a bad user experience.

(Amazon Web Services)

70% of online businesses that fail do so because of bad usability.

(Uxeria)

In short:

Having a slow database with a fierce competitor is like trying to pick up a box fight with a bear while having weighted mass on your ankles and your wrists.

You can't outrun it and you can't fight it!

But that's not all. Keep reading…

Who is faster: Lightning or Thunder? Light or sound?

#2 Downtime can kill your business in one blow! 🥊

March 10th 2021, a major fire took place in one of OVH's datacenters. 🔥🔥🔥

For some of OVH's customers, the downtime lasted up to 6 days!

Frederic from Lampe avenue was a victim of this crash.

He lost around 2 weeks of gross revenue and had to shut down 2 companies out of 5.

But he didn't have backups and he didn't reach out to me.

To stay in the fight, you should add a “nine” in your game

The 16th, Louis reached out to me.

We needed 3 days to recover from a SQL dump.

The cost of this incident was estimated to $5,600 a minute or $300,000 a day.

Louis lost “only” $65,812.

And he was lucky!

What are you going to do when the same happens to you?
Do you have a backup plan?
Do you have a failover?
How much time to recovery when there is a crash?
Did you ask these questions to your team?

In the real world, uptime can NEVER be 100%, it's always 99.?????%. The more nines there is, the more availability.

The real question is how close to 100% available are you?

Take a look at this and find where you are, if you can:

This was a rhetorical question: most small companies start with 98% maximum.

Let's be clear about one thing:

Crashes are like playing the Russian Roulette with a loaded gun: you don't know when and where it's going to fire. The only thing you can try to control is the number and the size of bullets in the weapon.

#3 You will lose data it's inevitable!

Louis, the client I worked with in the aftermath of OVH's outage event lost a small 523 lines in the ‘events’ table after recovery.

That doesn't look much.

He was relieved he could even recover the most part.

But some members of the team were making strange faces when they realized there were missing lines of data.

Their faces looked a bit like this

Service loss or data loss: Do you want to flip a coin?

Now, let's say you can recover.

Will you recover everything?

Every site page?

For every customer?

This is more unlikely than you realize. You WILL lose data. This is inevitable.

The right question is how much loss are you willing to accept?

Now, losing 523 lines of data doesn't seem so bad after all. But there is worse...

What were those 523 lines about?

Customer's items in their orders.

The problem was that the billing was done and the cart was paid for. But there was no trace of the items. What should they deliver?

No idea.

Victoria, the accountant took the decision to refund every customer: $24,622.84. Add that to the tab.

Have you kept you customers data in a different place? A different database?

OK then, but what happened to the rest of your customer's sales pages can happen to customers billing or contact data.

Can you imagine the consequences?

Double billing or no billing, losing client's billing informations...

A whole mess to pick up by hand.

But when you recover, the nightmare has just begun.

You thought it could not get any worse? Think again...

I told you it was just the beginning

#4 CorruPtion: A ~~tiC~~king time bomb!

There is a danger even more insidious. And it leaves a bitter taste.

It can happen when you recover your data, when your site operates on a normal day.

It can be obvious or silent.

And it can be a f***ing nightmare… It's called corruption.

Abdallah was victim of a bug from a bad version of database server system.

He defragmented the tables, but when he had finished he had sudden crashes.

Recreating the indexes stopped the crashes.

That was easy, but recovering the data proved more difficult.

We could not recover everything, this was a serious bug but we got more than 66% back.

This was one of my longest running missions I had to recover data by hands from corrupted server: It took months.

It could have been much worse. Not only you can lose data in the process but this can cause

more crashes,
more downtime,
more money lost,
more angry customers,
more depression...

It's a ticking time bomb! You never know when it's going to explode. You can never predict the amount of damage. It can hurt a few people or kill many and take the whole company with it.

But what if you had done everything right before to ensure this does not happen? Like using the latest version? What if there was a plan to make sure the risk goes closer to zero?

It's possible you could just get lucky and this never happens to you. But it's not over yet!

Last resort against corruption? You wish!

#5 Ready to pay a ransom yet?

Now that all is safe and your data lives happily ever after, what could happen now?

Ever heard of SQL injection?

DDoS attacks?

Ransomware?

These are the most widely known aspects of security breaches. The bad news is there are many others: exploits, backdoors, CVEs… Heard of Spectre and Meltdown?

They have cool nicknames, sure. But they could lead to catastrophic outcomes for you:

Losing customer's data.
Losing access to your own data.
Losing the trust of the user base.
That could damage your brand and reputation.

It could even be the end of your business. So have you covered everything?

A study from Varonis shows that up to 58% customers can leave a company after a data breach. It can take up to 5 years to regain the same level of trust and number of customers. This is especially true with younger companies.

Uber is a very good example: they hid their security breach for more than a year. It took them 3 years to regain the trust of the consumers after that.

Not securing your data is like going to the grocery store leaving the keys on the Lamborghini. No one in their right mind would do that!

The worst part is that choice gets taken from your hands once it has happened: you are at the mercy of the hackers.

June 2021, I had to tell Antoine his database security was a mess. No proper firewall rules, no passwords or outdated encryption for passwords. No secure connections, careless privileges. An outdated and unsupported database version! His website was not a major player with just 720K users, but the danger was still real. No wonder hackers have it easy. And the scariest part is that I come accross these issues with every customer!

Hope you can spare a few Bitcoins!

The eye of the storm sure does look like a safe place…

The calm before the storm...

Maybe you think you are safe and none of this will never happen to you.

That's what happened to all my clients before they called me.

Maybe you think you can find all the solutions yourself...

Allow me to save you some trouble then, and let's review all the options that won't work.

“Speed is a simple matter of more hardware”
Or is it? Configuration, tuning, rewriting code, switching protocol... Hardware is hardly the answer in 95% of cases.

“Making my own clustering solution (or taking a random one)”
Using a technology you don't master and you don't even understand? You're in for a lot of trouble. When the first node crashes and doesn't come back online, you will see how expensive that can be. Then you have to come up with another solution, which could be even worse.

“Snapshots!”
"There is nothing faster than snapshots for backups". True, but again, if it's done the wrong way, you might crash or corrupt your database.

“SQL dumps?”

They are fine as long as you don't have a big database and you know what you are doing. Did you ensure they were transactionnal? Is every vital element of the database in the dump with the right options? Are you sure they won't slow down the database? How long to recover? This is a lot of unknowns.

“What about HOT Backups?”

They are the best solution. But did you pick the right product? Some are slow, some are unreliable or outdated and some are down right very risky to use. And when you pick a good one, you need to be sure that there is nothing missing in your backup.

“I don't restore”

Did you test the backup by restoring? An untested backup is like Shrödinger's cat: you don't know if it's dead or alive till you open the box. Same goes with your backup: until you test it, it's both working and not working.

"Corruption? 1 chance in a million!"
Did you enable the right options in your instance to prevent corruption? Do you regularly check for corruption both logically and physically?

“It's secure enough!”
Are you up to date on the latest release? Did you enable the latest features? Are all access to your data secure enough? Are all passwords with expiration dates? Are they strong enough? Do you change them often enough?

Finally, did you make sure that all of this is on autopilot?

Anything could go wrong with a human intervention:

a bad copy/paste
an operation on a wrong server
a line forgotten in a code
a task forgotten on the TO-DO...

The list goes on. You just cannot rely on humans, they make mistakes!.

Life happens. It always does… eventually.

If you value your business, you cannot live on hope. You cannot think everything will work out for the best and nothing bad will ever happen. The only thing you can do is be prepared and have a solid plan. You have to take action now before it's too late. You cannot wait any longer!

Who Am I?

My name is Emeric TABAKHOFF. I am a database expert specialized in open source (PostgreSQL and MySQL/MariaDB). I help big and small companies secure their data and increase their sales with the fastest databases. I have helped dozens of clients get those results. I have encoutered all these issues with:

speed
availabitily
backups
security
corruption
migration

My job is to prepare, plan, implement and always be many steps ahead.

A few clients:

* EDF (General Electrics of France)
* Ministry of Europe and Foreign Affairs
* Virgin Mobile France/RED by SFR (mobile communication)
* Agirc-Arcco (Pension Funds)
* Monoprix (Retail)
* formerly Iko System (Marketing software)

* Animalis (Retail)
* Dating site (for confidentiality reasons I cannot disclose this client's identity)
* Traveljuice (flight scheduling)

I have worked on many different environments and very different situations:

Speed (SQL & PLSQL: triggers, functions, procedures)
High-availability
Clustering solutions
Cloud
Benchmarking
Hot Backups
Cold Backups
Migrations
Upgrades
Corruption
Scaling
Tuning
High-Writes environments
Load Balancing
Distributed Databases
Partitioning

Malt: 18,000€+ earned; global rating 5/5

Incoming payments from my clients

- GitHub profile: I am an active contributor of the PostgreSQL community.

SALT (for postgres) as a contributor

pgbench-tools (contributor)

maintenance_schema (creator)

…and the latest project: PostgresDBA (contributor).

This is my profile on Malt (freelance plateform):

Think of me as the “Ferrari F1 Mechanic” for Databases.

What now?

I have good news. It might look like there is a lot to cover but it's actually simpler when the plan is already written down. Even better when someone can do it for you. You don't need to worry about doing things yourself. You don't even have to find the person for the job. And you don't even have to tell the person WHAT to do!

I know exactly where to look:

X specific points for performance
Y points for availability
Z points for backup and restore
ZZ points for security
3 points for corruption detection

I will not disrupt the service or the database in any way.

I have even better news! It's not going to cost you 10,000€. All my database auditor friends charge between 1000€ and 3000€ just for… the database part. But it's not either 1,000€ or 100€. Far from it.

You can have me checking out:

your architecture
your application
your network
your server
your database
your data model
your queries

All for 1€ for a complete assessement. That's it. Simple as that.

What's the catch?

No catch, I just cannot do it for free. Why? Because, without this symbolic value, you cannot have a binding contract with me. This contract has a confidentiality agreement. A must have for a complete audit with sensitive data. Plus, my business is covered by an insurance from Hiscox (the best on the market) that can get you up to 5.000.000€ to cover the damages.

But I have to tell you I have never had to use it once in 10 years! And why would I risk my professionnal reputation?

I can safely say, you have very little to lose and a lot to gain.

But why just 1€?

Because I see a lot of businesses afraid of the cost of an audit. They just delay this formality until something catastrophic happens and they cannot wait any longer. This is what happened to Jerome, Louis, Antoine and Abdallah...

And I don't want this situation to happen to you or anybody. Losing data, availability or responsiveness should never happen to any business in this era. We are past 2020!

What if you could keep the speed up for months and years to come?

What if you could make your service available enough to your customers?

What if you could make sure that you could start right back after a crash or a human error?

What if you could make sure there is a very low probability of corruption?

What if you could make sure your data is secure?

What if you knew that you could call the right person to get you out of a bad recovery and get your data back any time of the day or night?

How much money could you save?

How many customers could you keep?

How many more could you acquire?

How much...?

Having doubts? That's perfectly normal

Any doubts?
If you are baffled by all these questions I think it's time to call someone.

There is no harm in taking somebody else's counsel. What is the risk of me taking a look at your architecture? You have nothing to lose. You are even protected by a contract with a confidentiality clause and an insurance. (For 1€, I am the one losing money here.)

Don't want to get more revenue from your website by keeping more customers for a very long time? It's fine by me, it does not change anything for me. You can always leave this page and this opportunity. I cannot have multiple clients at the same time anyway. And most important, I don't take just any client coming my way. I have to be a little picky. I only take clients that take action. Clients that won't leave anything to chance. Clients that won't leave any doubt in their mind.

If you are ready to call, you can book a slot right now so you can tell me all about your situation. This way I can know if we can work together. But I cannot promise there will be a spot left for you. Just book a call, and we'll see about that together.

Before I forget…

There is one more thing I haven't told you about yet: how I made the page slow on purpose!

See the photographs? I put in a lot of them on this page. And they are big (1MB to 7MB each). Imagine all your customers doing the same. Over and over again until this app is not usable anymore. The max size for pictures is set to 10MB. It's way too much! It should never exceed 100-200kB. For a page to load under 1 second, the best practice is to have a total page size of 500KB maximum. And don't count on compression on the fly from your server to get you out of this, it increases overhead on processors and memory. In the long run, it is going to become slower and slower. This is free advice.

And this is just a simple example.

Now click below to book a call. See you on the other side.