A number of articles surfaced last Friday about AWS selling network switching infrastructure used in its own environment to enterprises. This news had a ripple effect in the market as some of the major network companies saw a dip in their stock prices.
If the sources are accurate, AWS is, at the very least, contemplating this strategy. If it decides to enter the market, this could be highly disruptive to the status quo. Let’s take a minute to think about this, though--it’s not like AWS will be selling data center network switches online – well, at least not yet - I actually just checked and didn’t see any Prime Day specials. If they are seriously considering entering this market, it does highlight a number of things that AWS may have to do to capture the on-premises enterprise market share.
A few of those things may include:
- Sales: Building out an enterprise sales force or at least creating a strong set of channel partners in the areas in which they want to roll out offerings would take time. The fastest path would be to grab existing partners of the major network companies (Cisco, Juniper, Extreme, HPE, Dell EMC, Arista, etc.), but this would still take time. Established network companies spend millions building up a professional sales force (account execs and sales engineers, sometimes even specialist overlays) and having a direct presence is very important to enterprises, or at least it always has been. For example, when something goes wrong, I know many enterprise account execs that will be at the customer data center at 3AM helping to get the appropriate resources to fix the problem as soon as possible – this is what enterprises expect.
- Training & Implementation: AWS would have to create a training program and depending on the above sales path chosen, potentially an implementation arm (read: services) to deploy said switches and train customers on their use. In addition, the sales force, partners, and support personnel would need to be appropriately trained. It is clear that enterprises want to reduce their CapEx and OpEx, but still want to deploy solutions. At one of the industry shows this spring, there was a discussion about disaggregated OS and switches and all the benefits that would be derived from this arrangement, and then an enterprise customer stood up and said, “This all sounds great, but who is going to reaggregate it all for me?” Enterprises have typically wanted to consume solutions and need partners (vendors or VARs or SIs) to help accelerate the adoption.
- Support: Enterprises demand and expect 24x7 support. As I mentioned above, support is critical for enterprise clients. Knowledge boards and wiki-pages are great for perusing information at your own pace, but when your network is down, you probably want to speak to a level 3 engineer ASAP. Companies that excel in providing outstanding support (Extreme, Cisco, HPE, Juniper, Dell EMC, etc.) are able to leverage that as a competitive advantage when selling their products. AWS would need to build out world class support for deployed network switches.
- Roadmap: While the majority of use cases for a hyperscale environment would most likely be covered by AWS and its own use, I would expect that enterprises, especially in different verticals, might have use cases that would dictate additional features or functionality. I expect that AWS has a sufficient number of engineers to develop for all, but might it be a distraction or force them to prioritize the cloud over the enterprise in a pinch?
Does any of this prohibit AWS from entering the market? Not at all. AWS can address all of the above or take a completely different approach to selling – especially to digital-native companies (think less than 10 years old). It has already shaken up the status quo with cloud computing and can possibly do it again.
Does it impact the timing should they decide to join the networking fray? Most likely, unless of course they have been quietly doing all of the above and we just don’t know about it.
Does this mean it is a crazy idea? Not really. Thirty-six percent of respondents to ESG's Network Modernization research stated that they would like to emulate a cloud/hyperscale network environment. I expect that number will grow over time. One must keep in mind, however, that all the incumbent enterprise networking providers see this same data and are talking to these customers every day. As such, almost all have announced plans to disaggregate the Network OS and functionality from the hardware – and are at different levels of progress. Point is, none of the existing network providers are sitting on their hands waiting.
Lastly, it is clear AWS wants to get closer to the enterprise to help influence the cloud service used, much like Microsoft has been successfully doing with Azure Stack (choose your own infrastructure supplier). In fact, just last week, Azure quietly announced a Virtual WAN offering to help enterprises connect to the Azure cloud with SD-WAN solutions from Citrix and Riverbed tightly integrated to ensure a smooth transition. Note that AWS has also recognized the importance of SD-WAN and worked with ESG to create a report to demonstrate how easy it is to connect to AWS, but these do not have the same level of integration Azure has. Another example is the relationship Cisco and Google have to deliver tightly integrated solutions and AWS itself has a relationship with VMware to accelerate cloud adoption.
So as the summer heats up, so too does the network market. I look forward to seeing whether AWS ultimately decides that launching its networking solutions will be the best path to drive greater enterprise cloud adoption. Stay tuned…