What have we learnt from AWS ReInvent 2017?

AWS is innovative, it is fast-paced, it is dizzying in its ability to churn out new products at the rate of hours not weeks or months like much of its competition.

AWS ReInvent at Las Vegas this week is no exception, and (disclaimer) whilst I have not been physically there, I have been watching from afar with great interest.  

So here is my take on a few of the products that have been announced amongst the 100+ that have been released in just the first day.

Four things that really caught my eye and it seems a lot of other people:

Aurora Serverless. If you have an app that is well-written that needs the use of databases but don’t want to pay for dedicated instances then this is for you. It works well when workloads are unpredictable and it also charges on a second by second basis with the added feature that you don’t put in what you want in terms of database size up front - you put in a minimum and a ceiling. Potentially a great way to reduce costs compared to running dedicated instances as its dynamically auto-scales every 5 seconds. Even better, storage and processing are separate, so when you aren’t running workloads, you aren’t paying for instances. I like this....a lot.

DynamoDB Global Tables - admittedly, this is for the realms of large enterprises as the pricing will be prohibitive for anyone else. With this product, organisations can create tables that are automatically replicated across multiple AWS Regions. The big benefit is that you no longer have to manage the replication process and On-Demand Backup becomes more durable and available, this makes meeting some regulatory requirements easier to fulfil.

AWS Fargate - Putting aside the confusing name, Fargate is, in simple terms, EC2 for containers. This product significantly reduces (or even can eliminate) the need to deal with cluster management and makes using containers easier - which is in line with the where the technology is going. Not many people truly understand cluster management, I don’t, and it always seemed to me as one of things that is in the realm of being a dark art for those that like to punish themselves by doing jigsaws when the picture changes every time you lay down a piece. What further interests me is that AWS have already hinted heavily that they will be investing a lot in the development of this product by launching in Amazon EKS (Elastic Container service for Kubernetes) next year…watch this space.

S3 & Glacier Select - OK, so I am less impressed by the tech, but this is a product that is for me about 3 years late. An absolute must for any organisation that seriously uses AWS for its business. S3 Select enables applications to only retrieve the data subsets that it needs so you you get much better performance rather than the entire object. I think this great as this means an order of magnitude of performance improvement as you will consume less data. Glacier Select now allows you to retrieve data at different speeds (and obviously cost) which really helps those with regulatory demands and high amounts of data.

The three things I am really hoping from AWS now are:

Pricing - I would love to see AWS make its pricing model less complex and confusing. I have long said that the holy grail of the economics of cloud is to have an ability to accurately price a design in AWS, GCP and Azure. Unfortunately, I don’t see that changing in the short-term.

Lambda - It maybe is just me, but I am still yet to see a really cool, convincing story that proves the value of Lambda. I can see its value, I believe in its value, I just can’t prove it!!

AI and ML proof points - Again, we all know the value but unless AWS is working on something really big in the background or they have made a strategic decision not to go down this route (which I emphatically don’t believe), it seems that this is one area that the mighty yellow machine is losing ground to its competitors.

One last observation. I have always found that the greatest strength and the greatest weakness of AWS is the sheer size of its portfolio. With over 320,000 SKUs on its books, I have long harboured a fear that the complexity of its portfolio will make its harder for its customers to fulfil their business needs because they struggle to find what they want. That is not to say that AWS are releasing products that their customers don’t want - on the contrary, in most cases they do - but complexity does not make good bedfellows with customer purchasing.

What are your thoughts? - please feel free to comment and engage!