by: Mike Waas

Datometry Airlift: Moving From Oracle to PostgreSQL Without the Hassle of a Conventional Database Migration

Commercial databases are the single most expensive component in the IT stack. Oracle is no exception. So, it’s no surprise that customers have long sought cost-effective alternatives. Open-source databases like PostgreSQL have a big allure: they offer the full functionality of a mature database system without the license fee. 

However, database vendors have what is probably the strongest vendor lock-in of all of IT. Conventional database migrations are a mess. They require you to modify all existing applications in order to work with a different database. And, according to Gartner, most migrations fail. 

We here at Datometry have long recognized there’s a better way. What if the rewrites could be done in real-time without having to alter the application itself? Sounds too good to be true? We’ve pioneered software solutions that do exactly that. With Datometry, your applications can get the database functionality they need independent of the database you use. 

Today, we announce Datometry Airlift. With Airlift, we’re bringing the functionality we developed initially for Fortune 500 enterprises to a much broader audience. Database and IT leaders who look to move their business from Oracle to PostgreSQL can now sign up and be among the first to get their hands on this offering. 

How does database virtualization work? 

Virtualization has changed the face of computing in the past 30 years: from server and network virtualization to storage virtualization. Even Apple’s highly successful Rosetta is a kind of virtualization. In any case, the principle is the same: a software layer provides an abstraction, so applications are decoupled from the lower infrastructure layer. 

We’ve been bringing the same principle to databases now for about a decade: Database Virtualization (DBV) emulates one database with another. Practically, DBV makes PostgreSQL look and feel like Oracle so your applications originally written for Oracle don’t need to change. All functional discrepancies are reconciled in real-time. 

Say your application uses queries with quirks like ‘CONNECT BY’ — DBV turns them into ANSI syntax in real-time. You use Global Temporary Tables — DBV does the table management for you to create the fully equivalent functionality. And the list goes on. In short, if you can think of a way to rewrite it manually, DBV can do it for you automatically. 

So, what’s the architecture looking like? DBV sits between the application and the databases. It accepts requests from applications and translates (or emulates) statements and submits them to the database. On the way back, it converts the data into the format the application expects. 

Is this really possible? 

Well, we’ve been doing it for almost a decade. We have a great track record of virtualizing some of the world’s most sophisticated and demanding databases, at banks like ABN AMRO or retailers like ADEO. Our customers are Fortune 500 enterprises around the globe who use Datometry Hyper-Q to virtualize their mission-critical data warehouses. 

Just how good is it? Historically, our coverage is 99.5% or higher, on average. These numbers are in stark contrast to the typical 60%–80% of many migration tools out there. Best of all, our customers have transitioned over a single weekend without even telling their business users. By the time they arrived on Monday, the transition had already occurred seamlessly. 

So far, we’ve specialized on high-end enterprise database warehouses (EDW). These EDW systems run large mixed workloads that combine constant transactional churn and heavy analytical querying. They range from sub-second updaters to multi-minute aggregation queries. And now, we’re bringing this experience to a broader spectrum of workloads on AWS. 

What about performance? 

Of course, there is no free lunch. Remember when you first moved an application from a bare-metal server onto a VM? The flexibility and power of virtualization comes at the expense of a couple of percentage points in performance. For most everybody this is negligible — otherwise, the cloud wouldn’t be a $100bn business. 

In practice, the impact depends on your applications. If millisecond processing is your thing, virtualization may not be right for you. For the other 99%, virtualization is very much the solution. Our colleagues in server virtualization often cite 8% as the threshold to stay below. We think that’s a good target. 

Things aren’t that different with database virtualization: there are workloads for which the overhead of virtualization may not be acceptable — for the other 99% it changes how they do business. The trade-off between years of rewriting applications vs. virtualization is almost always in favor of virtualization. That’s been true for server virtualization as it is for DBV. 

What is Airlift? 

With Airlift, we’re bringing the success of our enterprise product now to all businesses. Airlift will be a marketplace offering for AWS that lets applications written for Oracle run as-is on PostgreSQL. Airlift is different from our enterprise offering in that it addresses the do-it-yourself segment of the database market, as opposed to the enterprise. 

Based on our initial market research, we chose AWS as the first platform for Airlift. As a destination, we selected PostgreSQL v14 or higher. We will also consider adding support for other flavors of PostgreSQL and cloud platforms, including managed offerings and third-party versions. 

Now, here is where you come in. If you’re an Oracle user or administrator and your business is looking to move to PostgreSQL, we want to hear from you! Let us know what features you need, and what your workloads are relying on. If you have tried and failed to migrate in the past we’re particularly interested in your story. 

We plan on releasing Datometry Airlift later in 2023 as a marketplace offering on AWS. Based on our experience, we’re confident we can create a compelling offer for most users on AWS. Then, going forward, we will be extending Airlift continuously. If we have not covered you with the initial release, we’ll get there soon. 

Ok, where can I sign up? 

Starting today, we’re accepting registrations for our early adopter program. By signing up for Airlift, you’ll receive monthly updates on the status of the project. Better still, you get a say in steering the fine-tuning of our offering. We even invite you to make the case for the prioritization of specific features. And, of course, you’ll be the first to learn about launch details. 

We’re looking forward to you joining us on this journey. We’re sure this is going to be worth your while! Welcome to Airlift — the countdown has started. 

About Mike Waas CEO

Mike Waas founded Datometry with the vision of redefining enterprise data management. In the past, Mike held key engineering positions at Microsoft, Amazon, Greenplum, EMC, and Pivotal. He earned an M.S. in Computer Science from the University of Passau, Germany, and a Ph.D. in Computer Science from the University of Amsterdam, The Netherlands. Mike has co-authored over 35 peer-reviewed publications and has 20+ patents on data management to his name.