Back in time with Amazon Web Services
As you know I come from a %100 systems background. I had the luck and suffering to pass the year 2000’s where we literally had no cloud computing. Everything was about building a data center or hosting your servers in one.
Spoiler: If you an inpatient go ahead and skip the end of the article!
Back in 2008 working for a tourism company we had the mission to create a new data center, new database cluster and get a better fiber connection to where we were, about 60 km outside of Barcelona.
The whole process began with experts visiting the two buildings, choosing a good spot, and start planning the creation. We had to be careful, as it was going to be the HQ where we would host all the servers for the company I worked for.
The architecture was done, then it was time to start working. The company hired a team to do the cabling, we found providers who could do us the cabling, then we had to go ahead and build a new system of backup power with gasoline which would kick-off when the lights went off.
After about 1 year and about around 1M Euros, we had the datacenter done. We had 2 connections of fiber coming from two different directions to be able to cover all the disaster recovery scenarios. The air conditioners were working great and the alarm system in the door let us in with a finder print.
That done now we had the mission to move our Oracle servers to a new cluster. For that, I worked with an external consulting company with whom we searched the best machines in the market which would hold the queries coming from our tomcat servers to server our clients asking for hotel allocations. Talked to the hardware provider and the prevision to get the hardware was about 1,5 months. We waited while we were trying to make sure the old system was up and running (yes 1 server in a forgotten closet in one of the old buildings). Passed 1,5 months and we had the servers. Now it was time to rack them up. I had to personally add ram modules myself to each server and ask for help while racking them up. And then came the installation of RAC system. Two clusters of Oracle servers with one server on the second building to do the backups with Dataguard.
We were not done yet. Then came the biggest surprise in licenses. We were asked to pay around 1M+ of licenses for the cluster and the Dataguard. And we were not done yet. We needed a system to monitor the system which was called Oracle Enterprise Manager. And you guessed with it came with a license which we had to pay.
Once the system was up it was now time to make sure the database was working correctly and was able to respond to queries without any performance issues, which ended up being a problem. Then it was time to add more servers to the cluster, which started the whole loop again waiting for servers, licenses…
We did it. After about 2 years of hard work, 1,5M Euros we were giving service. We had the 24/7 on-call which I was where I would receive alerts on the problems of requests building up and increasing the database queues up every night.
So why am I writing all these? You know those moments where you have things which you wished you did, you had in your early days? Well, this is one of the things that makes me wish I had Amazon Web Services in the old days.
Let’s barrow the time machine from Emmet Brown and go back to future with AWS.
Back in 2008 working for a tourism company we had the mission to create a new data center, new database cluster and get a better fiber connection to where we were, about 60 km outside of Barcelona.Which we didn't have to do thanks to AWS. We opened up our console and fired up a couple of EC2 instances. With the ELB configurations, we had the opportunity to create a load balancing system in minutes. We prepared one server and created an AMI which then we would fire up other EC2 instances from. After 1/2 hour we had the AMI created with Tomcat installed and with all the configurations. We then created 20 more identical servers. Then it was time to create the database. We set up Aurora which took us all 1/2 hour to configure. Then we didn’t need to worry about disaster recovery. with a click, we enabled Multi-AZ configuration which replicated the data to another data center in the background. Then we had to make sure we had a system against the logical corruptions in the database. In case something went wrong, a script failed in the deployment time and we had to revert back to the old configuration we just had to click and enable Aurora Backtrack which let us do the Point In Time recovery for our database.
We then signed up for AWS Enterprise Support which gave us 24/7 coverage and one person (TAM) to cover all our needs which saved us from talking to multiple consulting firms and helped us getting a personalized help.
The whole project took 1 week (Including enterprise support contract) and less than 30K.
We then, had coffee for 2 years knowing that our infrastructure was in good hands and we didn’t need to worry about our alerts.
I wish I had AWS back then.
