AWS Summit 2023

As a large user of AWS, advocate of Serverless and with a passion for Event Driven Architecture the AWS Summit has become a staple in my calendar. At least whilst I wait for my Re:Invent tickets to turn up.

Keynote

You can always rely on AWS Summit to pull out some extremely impressive numbers and contributions in cloud technologies and this year was no different. AWS were pleased to announce their successful execution of an $1.8 billion investment into services surrounding the UK, showing to their customers, Dunelm included, that we were in safe hands for whatever the future could throw our way.

Technology can often be considered as a large contributor towards unsustainable practices, though AWS are amongst the leaders to set this record straight boasting some impressive multiplier savings when compared to similar offerings in the market.

Whilst Technology appears in STEM, being in the industry we all know how fast technology moves. So it’s really exciting to here that AWS are heavily contributing towards the next generation of engineers. Who knows, some of those engineers could be found contributing to the Dunelm code base in years to come.

The largest part of the keynote was attributed toward Generative AI, something that is believed to have reached the required intersection for ease of adoption and power in the technology. It was super exciting to hear from Swami Sivasubramanian about AWS’s thoughts behind how this technology could begin to impact every sector. For Dunelm, we’re not a mass player in this space, yet. Though through the Swami’s showcase and the flexibility that we’ve come to expect from AWS systems. WATCH THIS SPACE!

Session One:

DAT301 | Migrating Deliveroo’s Dispatcher Service to Amazon DynamoDB

Migrating can always be quite a scary topic, and large scale food delivery platform Deliveroo were able to articulate the benefits of adopting DynamoDB from the previous PostGreSQL in 4 simple steps.

1. Duplicate?— enhance your application code to insert data into the new DynamoDB
2. Shadow Read?— Perform reads alongside your primary DB to compare structures and validate. Fixing issues along they way.
3. Roll out?— As confidence builds, incrementally increase the responses provided by the new DynamoDB
4. Clean up!

With a number of workloads on SQL backed data-stores with largely known access patterns. There’s a chance that these 4 simple steps may get used at Dunelm.

Session Two:

PAR-02 | Bridging the Gap Between Security and Development Teams in Less Than 7-minutes

Regulatory pressures and a multitude of acronyms appearing across the technology landscape often leaves the remediation a must for any organisation. Especially where a breech can have multifaceted implications, cost/brand/reputation.

Often a space where goals can conflict.

Engineer?— I’ve written the code to the best practices and I’m ready to ship it. But my security tooling isn’t optimal for my processes.

SecOps?— I’m positioned to ensure the platform is as secure and compliant. But I’m unaware of the finer details of features being shipped.

Trend Mirco have identified though experience that tooling is great, but a collaborative culture is better.

At Dunelm we talk a lot about shifting left. This session left me challenged on moving security practices left, not to the detriment of shipping features.

Session Three:

PAR-03 | Building A Log Strategy So You Can Sleep Better at Night

At Dunelm, I’d consider our logging to be a good enabler for tracking down those pesky issues. And this session largely confirmed that our strategy is heading in the right direction.

Though there were some additional takeaways worth exploring:

• Configuring alerts for large spikes in logs or matching error criteria is great, but configuring alerts for silence can be just as powerful
• Anomaly detection within logging software can provide an additional lens when viewing logs against expected traffic patterns created over days, or even weeks.
• PII. No one wants to see Personally Identifiable Information in logs but sometimes, just sometimes, with the correct care, attention and security permissions this could provide a huge insight into solving an issue and ensuring any impacted customer can be quickly resolved.

Session Four:

SER304 | Building low-code applications with Serverless workflows

A bit like marmite, some love it, some hate it! Step Functions in many businesses has seemed to be something to shy away from for one reason or another. Though this session proposed the ultimate challenge of only using Step Functions. In fact, to start with them!

With Step Functions now offering 000’s of SDK actions integrating with 00’s of AWS services, users have the ability to hand performance, security and scalability back over to AWS.

Some people fear the cost, though this session walked the audience through a real life cost reduction exercise through:

• Optimising state transitions
• Combining Standard & Express workflow
• Utilising intrinsic functions

To conclude, Step Functions has the ability to easily build some of the long lived patterns that have existed within the technology world for many years.

Final Thoughts…

Soo many notes, soo much to research. I’m not sure I quite know where to start, it’s all so fresh. I can say I’m looking forward to applying some of the learnings and new techniques presented today back at Dunelm.

Thanks

As with all large scale conferences, it doesn’t just appear. So thanks to all the speakers and staff working throughout the day to help up-skill us and keep us in the loop of what we can do to engineer performantly, securely, sustainably and cost effectively in the cloud.

And also, the incredibly intelligent teams behind each and every AWS service or partner service that gives each industry the opportunity to thrive in their sector.