No one doubts that the engineering machine at AWS continues to churn out new features and they more or less reflect what customers have been asking for. In general, the Day 1 announcements have been refreshing and exciting.
Things that are great/amazing category:
- Aurora Serverless — this ought to save boatload of money for well written apps that need a database but do not want to pay for dedicated instances
- AWS Fargate — Cluster mgmt is one of those topics that appeals to masochists, but frustrates mere mortals. By eliminating the need to deal with cluster managers, AWS makes using containers easier. Related, AWS EKS will help those that are still in the camp of “lift and shift” and wants to maintain some kind of control.
- DynamoDB Global tables — this is way cool for enterprises. I do suspect the pricing will be cost prohibitive for startups though.
- S3 and Glacier Select — what took you guys so long to do this? This is a no brainer and is usefu
In the category of “Jury is still out”
- Amazon SageMaker — sounds like a good promise, if they are able to deliver on it.
- AWS REK Video support — this is an add-on feature and is an year late compared to Google. Will have to see how well they can deliver on this. AWS REK has been disappointment so far.
- Kinesis Video Streams — this is a cool feature — if their ML works well, this can become very useful.
- Amazon Transcribe — a speech recognition service? A day late, may be even a dollar short. Will await to see what kind of adoption it will get.
- Amazon Translate- Has anyone checked the date? This is fricking 2017 and you are announcing a translation service? Someone at Amazon has been asleep at the wheels for way too long. LOL.
- Amazon comprehend — This will be a cool service if it can deliver on the promised capabilities. The ML models are a commodity these days, so its value is directly proportional to the amount of training. Perhaps, Amazon has an edge here, because they do have access to millions of books.
In the category of “Things I don’t understand:
- IoT Device Mgmt
- IoT Device Defender
- IoT analytics
I guess these are useful for those businesses that are investing in devices and have a real use case. I will refrain further comment on this.
In the category of “What I wanted to see, but didn’t”
- Pricing model — AWS pricing model is a Cluster Fck. Its a disgrace, complex, confusing, screws customers. I would have liked to see AWS invest some into reducing the complexity here. i would have liked them to kill Reserved Instances and offer sustained use discounts for every customer.
- Make it attractive for startups to continue to use AWS (pricing, pricing, pricing) — I get the sense that AWS wants enterprise business and they want to drive away startups. This could prove dangerous to AWS.
- Lambda proof points — I would have liked to see more customer success stories with Lambda.
- AI/ML success stories — I left the keynote thinking AWS is struggling mightily in this space. The keynote did nothing to change the opinion of AWS as lagging behind other cloud vendors.
I am left with one feeling. This might be the first year of Re:Invent keynote where I felt that perhaps AWS is showing some early signs of behaving like a legacy vendor. I doubt that they will become mediocre or a legacy vendor anytime soon, but they seem to be thrashing a bit when it comes to pushing the envelope. The keynote started very strong, but ended on a weak note.
Thats all folks. Thoughts? Share here or on Twitter.