AWS does a quarterly housecleaning/”Googling” of its services, and has come up with what appears at first glance to be a surprisingly long list. However, going through them put my mind at ease, and I hope this post can do the same for you.
what did you get the axe?
19 services have been shut down (“maintenance mode”), four are being retired (“you can’t use them after the upcoming date”), and one is being desupported (“it’s finally finished”).
Some are worrisome: Something like “Cloud Directory” seems like it would be difficult to change, until you think about it and realize you never use it. now when you In fact Think about it, you don’t know anyone who has that.
The ones that really stood out to me are “Amazon Glacier,” “S3 Object Lambda,” “Snowball Edge,” and “CodeCatalyst.”
those that matter
Glacier is a red herring. At one time Glacier had its own service, with its own API. Now, this is an S3 storage class. What they are doing is removing the ability to interact with Glacier through their own API, which has always been extremely annoying to work with.
S3 Object Lambda has always been a bit strange. You can still operate Lambda on S3, and are at least likely to see service improvements in actual Lambda; Object Lambda has been dead for years.
It was a huge accomplishment when CodeCatalyst launched, and nothing was heard from customers or AWS after that. It could have been something, but the desire to make it that thing has apparently caused AWS to depart with some of its better talent.
That leaves Snowball Edge. This is a strange thing, because a bunch of customers have local EC2 instances running on them, as well as using them for data transport jobs. Those clients can continue to do this (for now, at least), but if you’re building something new that takes advantage of it I would suggest making other plans.
everything else
A bunch of modernization stuff that’s been searched for on Google has just been pulled into AWS Transform. New service marketing, similar capabilities, and what’s more, if you’re migrating you’ll want to at least aspirationally think that you won’t be doing it forever; End your damned stay already.
IoT Greengrass V1 lets you run Lambdas on your gear, and v2 has been available for several years. I question this a bit, as it would require updating things deployed in the client area, but… if it’s running completely separate and hasn’t been updated in such a long time, then go ahead, I guess?
System Manager Change Manager and System Manager Incident Manager are being phased out, with reasons ranging from “other System Manager capabilities with equally bad names” to “do what sane people do and use the best in class third party option instead.”
bottom line
Most of these criticisms seem to me to be the rotten fruit of AWS’s “launch a new service to solve problem X” approach that continued for too long. It was clear that not all of these would be commercially successful, and I’m optimistic that cleaning up their shambles will give Amazon a chance to put more effort into the fewer things that actually matter to customers.