Charting Stacks

AWS re:Invent: Hybrid, Greengrass and the Future of On Premises Hardware

Share via Twitter Share via Facebook Share via Linkedin Share via Reddit

The headline grabbing announcement for many people at re:Invent was AWS Snowmobile. A container full of storage and compute to help people migrate into the cloud. As my colleague Stephen O’Grady noted during our re:Invent recap, data is the ultimate jobs to be done barrier for Amazon.


While Snowmobile is the eye-catching announcement, it is well worth looking at a number of the other key edge and hybrid announcements made during re:Invent and tying them all together.

The Edge, The Future & AWS Greengrass

During the day one keynote Andy Jassy spent some time talking about the future of on premises computing, which he essentially boiled down to being edge devices such as sensors, control systems and so forth.

This is, of course, very consistent with Amazons vision that ultimately everything ends up in the cloud, but the direct acknowledgement that customers want to do significantly more work on the edge is important and matches the general trend towards edge hubs and processing that we see from other vendors such as Microsoft.

Firstly, lets delve a little into AWS Greengrass. At its core Greengrass allows you to AWS Lambda functions on edge devices, with or without connectivity, but with all the functionality that connecting that AWS IoT brings. In short it is extremely powerful.

The existing model, far more prevalent in consumer IoT than in any other area, of IoT data going into the cloud to be processed has been broken for a while. Greengrass addresses this in a manner that is simple for developers to access and uses a paradigm that many are already familiar with.

This extends further than just IoT though, think of a point of sale system – a classic edge device. Combine an on premises point of sale system (despite everything you hear, many large retailers have point of sale systems that are completely disconnected from everything) with AWS Snowball for local storage and compute, and AWS Greengrass for processing and an S3 endpoint for more pertinent data to move to the cloud – you can see where this pattern leads.

AWS Snowball Update

Amazon announced an update to their AWS Snowball offering, with additional storage capacity, on-board compute capabilities and clustering of multiple devices. We have touched on the compute aspect already in terms of AWS Greengrass – the clustering capability is also worth looking at in more detail. You can now cluster up to five Snowballs together, but more importantly as the devices fill you can remove them from your cluster, send them back to AWS and replace with a new device.

Now think about this in terms of long term storage as an example. Imagine you are in the not uncommon situation of having large volumes of data that you need to store for regulatory reasons, but you only need, at most, occasional access, once an initial action has been taken upon it.

Instead of maintaining your own large storage setup, back up facility and so forth you instead rotate a cluster of Snowballs and drop your data off into Amazon Glacier as required. Such a simple setup changes the economics and design patterns around storage architectures dramatically. Aspects of data gravity become far less relevant.

If the volume pricing is set correctly we could see AWS Snowball and AWS Snowmobile the storage significantly disrupting the entire storage market in the next eighteen months.

VMWare and Amazon


We have had many conversations around the VMWare and Amazon announcement since attending VMWorld Barcelona earlier this year. The overall feedback from customers has been positive. This is hardly surprising, as we have written before, VMWare provide an enterprise on-ramp for technologies such as containers, this is equally applicable to moving into cloud, specifically AWS in this case.

However, the details are still being fleshed out. We know you can ‘bring your own license’ to the VMWare on AWS party, and your management viewpoint will be via vSphere rather than the AWS interfaces. From Amazons perspective, the VMWare deal is very straight forward, it gets more workloads onto Amazon and removes a reasonably significant customer friction point.

From a VMware point of view this future proofs revenue in the medium term while customers make a transition to AWS. The impact on their partner ecosystem and longer term revenue growth is less clear, as is the positioning of some of their newer products such as NSX running on AWS.


AWS are confident that the future of cloud is hybrid, it is their definition of hybrid that is somewhat different to all the other vendors. But the combination of offerings raises some truly compelling possibilities.

Disclosures: Amazon paid my T&E for re:Invent. Amazon and Microsoft are current RedMonk clients.

No Comments

Leave a Reply

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