Cloud vendors compete based on their services, which creates lock in. Independent software vendors compete based on providing the same user experience across multiple cloud providers, which breaks the lock in.
Of course, companies using a single cloud provider may not be very concerned with lock in, but at some point it may become important to move an application from one cloud provider to another, for financial or other reasons.
Independent software companies delivering compatible software across cloud providers reduces the impact of cloud provider lock in, and perhaps eventually will eliminate it altogether.
Let’s look at industry history for precedent, and the database market for a specific example.
Minicomputer Vendor History
What happened in the minicomputer market is very similar to what’s happening in the cloud market today. Minicomputer vendors all had their own proprietary hardware and operating systems and started to develop more and more proprietary software as their customers kept asking for more and more capabilities, and competed with each other on the basis of differentiated hardware and software.
When I was working in the commercial software division of minicomputer vendor Digital Equipment Corp (DEC) in the mid-90s, I used to ask why we were investing so much in software when the company was primarily interested in selling hardware.
The simple answer was that our customers were asking for it – they bought DEC minicomputers because they were cheaper than mainframes, initially for one or two specific applications, but the longer they owned them the more applications they wanted to develop for them, and the more software products they wanted to buy in order to do so.
They wanted to (or actually in most cases had to) buy not only the hardware and the operating system that ran on it from the same vendor, but also wanted to buy the database, office automation, UI products, and transaction processing monitor, etc. from the same vendor.
It was the same for the other major minicomputer vendors: Prime, Wang, Data General, HP, Honeytwell, and TI. Each had their own proprietary hardware (e.g. CPU chips), operating systems, and application software such as programming languages and databases.
Eventually standard CPU chips from Intel took over the processor market (for the most part) and Unix (especially Linux) took over as the market leading operating system for these machines, resulting in the death of the minicomputer.
In particular, however, long before that happened, the rise of the independent software vendor, with the popularity of Oracle database as a leading indicator, started to break the lock-in created by minicomputer vendors selling their own proprietary databases.
Cloud Providers
It’s been pretty much the same with cloud providers. AWS started with EC2 (compute) and S3 (storage) in 2006 but as more and more applications went to the cloud, customers started asking for more and more services. It was the same for Azure and Google Cloud. (And no doubt now starting to happen for the Oracle Cloud.)
And just like the minicomputer vendors, cloud providers compete with each other on the basis of these services, differentiating themselves by the variety, performance, functionality, and quality of the services they offer.
However, just like the minicomputer vendors, going with one of the cloud provider’s service offerings means getting locked in to their platform, since migrating to another cloud provider would involve significant work.
Last year at the HPTS conference, James Hamilton, SVP and Distinguished Engineer at AWS, described in his keynote address the trend of cloud providers toward specialized hardware and integrated software as a kind of return to the minicomputer days.
Page 5 of his presentation references a minicomputer TPC-A benchmark from 1995 performed using DEC’s database Rdb (recently sold to Oracle), TP monitor ACMS, and VMS operating system running on the proprietary Alpha CPU. This is an example of how proprietary systems used to compete, in this case on the differentiation of achieving the best results on a performance benchmark.
From page 13 on James talks about the recent AWS investment into custom servers and chips, for example for training AI models and networking speed. Again for competing with other cloud providers on the basis of performance. Back to the future of the minicomputer days of vertically integrated hardware and software systems
Database market
So let’s take a look at what’s happening in the database market, where we are seeing all kinds of innovation and rapid evolution, especially for AI model training and real time analytics.
The convergence of time series and real time analytic databases is one trend, as is the emergence of vector databases for training AI models on unstructured data.
Oracle is promoting the converged database trend, which MongoDB seems to be echoing, along with others promoting the same data store for transactions and BI (one database for everything?).
Cloud providers continue to offer multiple database services for a variety of use cases.
Page 10 of James’s presentation notes 13 database offerings from AWS alone. For example AWS is offering DynamoDB for NoSQL and Aurora for SQL
Among Microsoft Azure’s multiple database offerings are Azure Cosmos for NoSQL and SQL, and Azure SQL for SQL.
Google Cloud’s multiple database offerings include Bigtable for key-value (NoSQL) and Spanner and Cloud SQL for SQL.
These and others of their various database offerings are differentiated for competitive reasons, but many of the offerings expose different programming models, and different functionality.
Customers have to be careful to avoid features and capabilities that are specific to a given cloud provider database if they want to avoid lock in.
Independent Software Vendors and Multi Cloud
Just as Oracle and other independent software providers broke into the minicomputer market and started offering the same product across multiple vendor platforms, we are starting to see the same thing happen in the cloud with independent database software vendors such as Snowflake, MongoDB, Couchbase, CockroachDB, MariaDB, and others offering the same product across multiple cloud providers.
Cloud providers continue to present challenges to customers who use multiple clouds, even when using the same database, however. Embedded services for security, monitoring, and alerting can be incompatible across cloud providers. Independent software providers are emerging in the monitoring, security, and observability markets as well, however, so it may just be a matter of time before these challenges are met as well.
This trend is likely to continue as more and more enterprises continue to move their applications to the cloud, and use more than one provider, eventually resulting in a comprehensive, standard set of services from independent software vendors across all cloud providers, eliminating lock-in (for those who choose to do so, at least).
Some more history here
People often ask why system software such as databases are being reinvented for the cloud, or why applications have to be re-engineered into microservices to obtain the benefits of cloud computing.
In other words, why is all of this happening in the first place – why are cloud providers reinventing or re-engineering basic software systems such as databases, and why do organizations have to face the migration and lock in challenges.
It’s a simple explanation: cloud computing infrastructure is different from traditional enterprise computing infrastructure, which means new software and applications have to be built from the ground up to get the best results.
The computer industry has seen this before, and not just once but at least twice (depending where you start), in migrating system software and applications:
- Mainframes to
- Minicomputers and distributed systems to
- PCs (and all) to
- Cloud
The success of each of these types of computing platforms resulted from their offering new and cheaper types of hardware and software, but in so doing created incompatibilities with systems and applications built for the previous type of computing platform.
I sometimes call cloud infrastructure “mainframe performance for a PC price,” basically because cloud infrastructure is constructed out of consumer components with high failure rates, but features software that delivers a mainframe level of reliability and performance.
In other words, cloud infrastructure is a highly reliable system composed of unreliable parts.To get the best results out of it you have to develop software specifically for it. Initially this has resulted in vendor lock in, as each provider has gone about this in different ways.
In the longer term, however, just as in the minicomputer market, independent software vendors are very likely to become the force that resolves this, and breaks lock in.
The Intellyx Take
The trend of the computer industry has always been toward standardization, from mainframes, through minicomputers, PCs, open source, and now in the cloud. This history indicates that cloud services will standardize as the market matures, with the independent software vendors leading the way.
Cloud providers are already starting to embrace the standardization in some ways, for example providing SQL interfaces for their databases, or the MongoDB interfaces for AWS Document DB and Azure Cosmos.
In the meantime, If cloud vendor lock-in is a concern, consider using software from an independent provider instead of from a cloud provider, such as databases, messaging systems, or observability tools.
Independent vendors have a different motivation from cloud providers: most want to sell their products regardless of which cloud (or clouds) they run on. They are not competing on the basis of differentiating from each other as cloud providers, which is what tends to create the lock in challenges of moving applications that depend on cloud provider specific services from one cloud provider to another.
As the cloud market matures, the cloud providers are likely to follow the lead of the independent vendors (especially if they continue to attract more customers) and standardize more services across clouds.
Copyright © Intellyx LLC. Intellyx is an industry analysis and advisory firm focused on enterprise digital transformation. Covering every angle of enterprise IT from mainframes to artificial intelligence, our broad focus across technologies allows business executives and IT professionals to connect the dots among disruptive trends. As of the time of writing, none of the organizations mentioned in this article is an Intellyx customer. No AI was used to write this article. Image credit: Wikimedia Commons, Scaramanga731.
