Beware of the legacy public cloud
Legacy is a scary and bad word in the computing industry. Most enterprises have “legacy” technologies, and all who do wish that they did not.
Now everyone thinks of public clouds as the next greatest thing, which means they cannot possibly be “legacy.” While the major public cloud vendors (Amazon, Microsoft, Google) currently drive a great deal of innovation, they might also be trapping themselves and their customers into legacy situations.
So, there is a real risk that the current public cloud leaders (Amazon Web Services, Microsoft Azure and Google Cloud Platform) could become legacy vendors. To understand this risk, we need to understand what creates legacy technologies and vendors and analyze the strategies of the public cloud vendors.
What defines a legacy technology?
To sort out whether or not it might be possible for the current leaders in the public cloud market to become legacy vendors, we must identify what creates a legacy vendor or technology:
- Lack of innovation: Legacy technologies cannot and have not kept up with the pace of innovation. For example operating systems such as Z/OS, Z/VM (mainframe), IBM iSeries (AS/400), and proprietary UNIX (AIX, HP UX, Solaris, SunOS, etc.) are legacy technologies.
- Ecosystem support: Legacy products and technologies lose their third-party ecosystems. This is a corollary to the innovation point above. Legacy platforms do not get the third-party developer support and the support of vendors building other kinds of complementary products required to sustain their market position.
- The vendor as the sole source of innovation and support: Legacy technologies rely upon a single vendor to support them and advance their cause. The best examples are IBM for their mainframe and midrange offerings; Oracle for their database; and IBM, BMC, CA and Micro-Focus for their enterprise management offerings.
- Proprietary: With one vendor as the sole source of progress, legacy technologies inevitably become more and more proprietary over time.
- Customer lock-in: Proprietary technologies tend to characterized by being difficult and expensive to move away from.
- Customer abandonment: A technology or a product certainly becomes a legacy product or technology when customers start to abandon it or wish they could abandon it.
Current state of the public cloud industry
If we consider AWS, Microsoft Azure and Google Cloud Platform as representative of the current public cloud computing industry, what we see is the following picture:
- We see the cloud vendors innovating at a very rapid pace in their own product lines. Amazon is particularly adept at this—rolling our new services on a rapid and continuous basis.
- In the course of rolling out these new services, the cloud vendors end up competing with vendors that provide those services. Examples:
- The Amazon Relational Database Service competes directly with on-premise instances of MySQL, Microsoft SQL Server and Oracle.
- The Amazon Elastic MapReduce competes directly with on-premise Hadoop distributions like those from Cloudera and Hortonworks.
- Amazon CloudWatch competes with the entire on-premise systems management and monitoring industry. Cisco just acquired market-leading application perforamance vendor AppDynamics, and Amazon appears to be rollout out a competing offering that puts Amazon and Cisco into competition with each other.
- Amazon Simple Queue Service competes with message brokers such as RabbitMQ and message buses like Kafka.
- The cloud vendors are also innovating in infrastructure. In fact, Amazon is engineering and deploying its own custom hardware and networking.
The legacy public cloud scenario
At this point in time, there is no evidence that the three major public cloud offerings are legacy offerings or are on a path to become legacy offerings. However, the following risks need to be watched very closely:
- By choosing to compete with much of the rest of the hardware and software industries, Amazon and the other public cloud vendors put a tremendous burden upon themselves to out-innovate those industries.
- By choosing to compete at all layers of the stack with both one another and standard enterprise computing offerings, the public cloud vendors risk becoming proprietary islands of innovation.
- There already are issues with workload portability in and out public clouds—even if you use them as simple Infrastructure as a Service (IaaS) offerings. If you take advantage of any of the higher-level services, you lock yourself into the APIs associated with those services.
- There are already clearly two schools of thought regarding the use of public clouds by enterprises. One school of thought is to move things to public clouds and to leave them there. The other is that starting things in the public cloud (dev and test) is fine, but for reasons of technical lock-in and economics, production workloads belong in a private/hybrid cloud.
Today the offerings from the three major public cloud vendors are at the tip of the spear in terms of innovation. However, by innovating across the entire stack, including application services, systems services, networking services and hardware, the three major cloud offerings risk diverging from enterprise hardware and software standards into three proprietary islands. To the extent to which the three public clouds diverge into their own stacks and customers become locked into these stacks, the path to a legacy scenario will be embarked upon.