AWS Relational Database Service for Db2

IBM, the Death Star, and rise of the Resistance.

In the spring of 2018, Ember and I attended a dinner with members of the Db2 Technical Advisory Board (TAB). This was a collection of respected Db2 end-users, clients, and consultants that had just spent a solid eight hours with IBM providing feedback on Db2. We were taking our seats at the table with some of the IBMers we had just questioned and challenged during today’s session.

I remember that evening well because of an unexpected conversation with an IBMer regarding our experience with the Amazon Relational Database Service (RDS) product. He wanted our thoughts on Db2 in the cloud, our experience with AWS RDS, how they differ, and what we would think of Db2 as an RDS offering. Once the IBMer excused himself for the evening, I turned to Ember with a confused look on my face.

Mike: “Whoa! Did IBM just ask us about our thoughts on RDS!? Members of the TAB have been advocating for that at the top of their lungs for years”.

Ember: “Not only have I been told no. I’ve have been politely told ‘Hell No, and please stop asking for it’”.

It was around that time that I noticed the start of an internal change within IBM. Forgive the bad Star Wars reference, but I had always viewed IBM as the evil Empire. They were big, powerful, wide reaching, and saw their products as the only end-to-end solution (especially in the cloud). Why consider anything else? Now I was starting to see a Resistance faction come to light within IBM. This faction believed IBM products were often the best choice, but not the only choice for a particular solution. This small band of misfits understood a solution didn’t always have to be a homogenous solution of IBM products.

Over the next few years, I observed a changing of the guard in IBM culture and strategy. This included an announcement in May 2022 of a strategic collaboration between IBM and AWS to deliver IBM Software-as-a Service. It seemed as if the Resistance was gaining momentum.

AWS Relational Database Service for Db2

Today you are seeing the first fruits of that partnership as IBM proudly announces the Amazon Relational Database Service (RDS) for Db2 at Amazon Re:invent 2023. (You can read Amazon’s official announcement here – Getting started with the new Amazon RDS for Db2).

Luckily, I was able to participate in the process of bringing Db2 to AWS. I sat in multiple customer feedback panels as IBM and AWS gathered requirements for the project. More importantly, I was able to participate in a multi-week beta program where I was able to get hands on with AWS’s Relational Database Service for Db2. I was able to poke, prod, break, load test, and generally kick the tires to make sure this product was solid when it launched.

To put it bluntly, I was impressed.

I had a small advantage when I went into the beta test for this product. I had worked with Amazon Relational Database Service for SQL Server in the past. Our shop had been early adopters of AWS friendly Db2 features (i.e. logging/backups to S3). I also had just become a certified AWS Cloud Practitioner. I knew how the various products played together, I knew some limitations going in, and I already had ideas on how I could “break” the product. But it was my experience with RDS for SQL Server that made me nervous for this beta test. In my opinion, there was a decent difference between an on-premises install of SQL Server and its RDS counterpart. Administration took a different approach. Features I was used to seeing may not exist, and many knobs I would use for tuning were gone. So, I was excited to see this product, but I was concerned about what I would find. Would it be a neutered version of Db2?

I spent six weeks building, migrating, testing, tearing down, and rebuilding to see how this product performed. My goal was to find out what it could do, find any limitations or concerns, and see if RDS for Db2 had a natural fit in our shop. I won’t lie, by the end of the second week I was becoming concerned at some of the limitations or bugs I had found. But that is when Amazon really stepped up and implemented a tight feedback and development loop. Problems were repaired and closed almost as fast as I found them. Limitations were reviewed and workarounds found. I was very happy with the results I observed after completing a load test. By the end of the beta testing window, I had discovered a solid product that performed well under load.

My biggest take away from testing was the “culture shock” a traditional Db2 database engineer may experience when trying RDS for the first time. For example, traditional Db2 DBAs are used to having the ability to SSH into a target server so they can work locally. You lose this ability with RDS for Db2, so many administrative commands are executed remotely via an instance attachment or database connection. A few of these administrative commands have been converted to Amazon specific Db2 stored procedures as well.

There are a few other oddities in the transition. You will need to rethink your backup, recovery, and high availability approach as the burden shifts to AWS through snapshots and multi-az deployments. Need to make a configuration change? Well… that is done a little differently now too.

It sounds a little odd or even overwhelming but trust me, it isn’t. The shift is not difficult or complicated, it’s just a change in your thinking and approach. To be honest, I was surprised how fast I was able to adopt this this shift in thinking and technology.

Over the next few weeks, I plan on writing a few articles to explain differences and help ease the transition from a more traditional build to RDS for Db2. However, if you are one of the lucky ones attending AWS Re:Invent this week, make sure to keep an eye out for Db2 sessions that will unlock in the Re:Invent conference catalog on Monday morning. Try to reserve a seat for the “AWS RDS for DB2” breakout session on Wed, the 29th, at 8:30am.

In the end, I only see this as a positive turn for IBM. Shifting strategic direction and corporate culture is not a quick process for a monster like IBM. In many ways, it’s like turning a battleship. She may not turn on a dime, but dear God you want to get out of her way when she finally gets pointed in a direction and gathers a head of steam. This product is a result of that turn and has been a long time coming. Welcome to the party IBM, we are glad you are here.

Michael Krafick
Michael Krafick

Michael Krafick is an aspiring Zamboni driver and well known twig-eater. During the day he is a Sr. Database Engineer for a Fortune 500 company. He is a frequent contributor to datageek.blog, an IBM champion, member of the DB2 Technical Advisory Board, and IDUG speaker Hall of Fame inductee. With over 18 years experience in DB2 for LUW, Mike works hard to educate and mentor others.

Articles: 34

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.