This is a guest post by Phil Shih, Managing Director and Founder of Structure Research.
AWS unveiled a new infrastructure deployment model – AWS Local Zone – that brings compute and storage infrastructure closer to end-users in a given location that AWS does not have an available region in. The first AWS Local Zone will be in Los Angeles, CA.
Details: AWS Local Zone is a new infrastructure deployment model that is basically an edge data center. The edge location will reside outside the regions where AWS currently has availability zones. The idea is to get localized infrastructure closer to end users that require very low levels of latency (in the single-digit milliseconds). The first Local Zone location is in Los Angeles and AWS is primarily targeting the multi-player gaming, video editing and rendering, film, content creation and machine learning industries. Compute, storage and database infrastructure will be available in this location along with other select AWS tools and services. Importantly, end-users will be able to connect back to the closest (or any) AWS region (in the case of this first Los Angles location that would be Oregon) and get access to the full portfolio of capabilities that AWS has. End users can connect an AWS VPC directly to a Local Zone.
Location: The choice of Los Angeles is an interesting one but speaks volumes about the new offering. LA has a critical mass of content and media vendors and is a gaming hotbed. The volume of workloads in this market demanded localized infrastructure that is highly performant and AWS simply could not do this reasonably serving out of its region in Oregon. To serve this use case it had to get out to the edge.
Use case and value proposition: This is pretty straightforward. AWS is looking to serve organizations that have previously used local data centers (colocation or on-premise) or cloud services because of the desirable latency levels with localized infrastructure. They would also be accessing different APIs and tools to run this infrastructure. Running on Local Zone gives it access to the same API. It also provides an OpEx versus CapEx choice (and benefit).
An extension to cloud just as the edge is meant to be: AWS Local Zone is an edge play that underscores the very important point: the edge is an extension and value-add to hyperscsale cloud infrastructure. Do not believe any hype that talks about edge eating the cloud or replacing the cloud. As this use case plainly demonstrates, they can and will be reliant on each other and work in tandem.
The infrastructure itself: AWS did not disclose a ton of details, but this launch coincides with the general availability of AWS Outposts (basically the native AWS cloud platform installed and pre-racked on proprietary hardware and available to customers as if it was an appliance). How do they relate? AWS Local Zone consists of Outposts-based infrastructure. Local Zone is basically dozens of these Outposts racks in a data center. The word at re:Invent is that these Local Zone deployments are going to consist in size of 50-100 Outposts racks. Outposts was largely designed for the on-premise data center or colocation. Local Zone is for customers that don’t have data centers or want to shut them down entirely. AWS continues to go after more digital infrastructure real estate.
Colocation makes sense: Another tidbit we came across: the AWS Local Zone is housed in a third party colocation data center. It would make sense that AWS uses colocation. It can’t get economies of scale by building (especially at this size) and it certainly can’t do it at a whim in any major market. Colocation is the perfect OpEx-based way to procure just enough capacity while having the ability to scale and grow that footprint as needed. If AWS needs multiple Local Zones in Los Angeles (it most certainly will over time), it would make sense to use colocation again. It is not just the scalability, but the time to market.
Angle: Well, it did not take long for AWS to make a direct edge play and it didn’t even bother to mention the word edge in the press release or accompanying materials (we haven’t yet reviewed Andy Jassy’s keynote in detail, but he may well have kept to script and not uttered the word). The AWS cloud footprint is highly centralized. It has been able to serve a large number of the scenarios out there … but not all of them. To get at the higher-performance use cases, like the ones prevalent in LA, AWS had to get out to the edge. This move is a disruptive one that is in line with the AWS MO: it does what it does well and methodically (raw storage, then scale-out compute, then mission-critical compute), proves the case and scales it, and then goes after more use cases. As it does that, through data gravity and proprietary tools and APIs, it locks down end users. The upside potential of Local Zone is the speed at which it can deploy and scale. AWS can spin up Local Zones pretty much as fast as it can procure colocation. It is going to expand this to multiple locations next year and even in major metros where it has a hyperscale cloud region, it could easily extend to the edge in these markets as well. We talked about the potential of edge for independent operators at this year’s infra // STRUCTURE event in Toronto. Outposts was a potential threat we identified and it looks like AWS is already trying to jump in and push the envelope.
About the author:
Phil Shih is Managing Director and Founder of Structure Research, an independent research firm focused on the cloud, hosting and data center infrastructure service provider markets. (more)
DISCLAIMER: The views expressed in this contributed post are that of the author, and don’t necessarily reflect the views of EdgeIR.com. Contact us if you want to contribute a guest post.
access edge | Amazon Web Services | cloud computing | edge servers | Structure Research