Almost 30 years ago, Java first appeared as a programming language. Its promise was Write Once Run Anywhere. It was an immediate hit with developers and created a huge market across different environments. That continues today, with GitHub ranking it as the third most popular programming language in 2022.
Since acquiring Java, Oracle has been the commercial owner of the Java Enterprise Edition. However, Java is firmly rooted in open-source and Oracle has continued to make that available. It has allowed a lot of companies to build their own versions of Java. Among those is Azul, which has the second-largest Java engineering team after Oracle. It also has a market capitalisation of over US$850 million.
Enterprise Times editor Ian Murphy sat down with Scott Sellers, President, CEO and Co-Founder of Azul, to talk about the company and the state of Java. Along with his co-founder, Gil Tene, Sellers founded Azul in 2002 to provide an alternative supplier for Java. Importantly, one that could address some of the deficiencies in Java, such as performance and garbage collection.
Azul started with hardware products
Talking about the early days, Sellers said, “Our first products were hardware products. We built our own Java microprocessors, and raised a tonne of money from venture capitalists. We sold several hundred million dollars of those products, so we did pretty well.”
As Sellers went on to say, the hardware market is a cash-rich game, so in 2008, the company moved out of hardware and into software. That mirrored what it was being asked for by customers. “We had heard from our customers at the time, we love you guys, we love what you’re doing a lot of great technology here. Can’t you just sell us a piece of software?”
Open-Source was a game-changer
One key factor for Azul was the move by Sun to create the Open JDK project and open-source Java. It benefitted the whole Java community and allowed more and more companies to build solutions around Java. It also opened up the opportunity for companies to create a business model around serving the fast-growing Java community.
As Sellers points out, it showed companies that they could take an open-source product and build an enterprise product console. It created a whole new business model. Interestingly, when Oracle acquired Sun and, therefore, Java, one of the conditions was that it continued to support and develop the open-source project.
Sellers commented, “They deserve credit. They’ve continued to make Java not only completely open source but, frankly, have really embraced it. In the case of Java, they have found a very good balance between serving the Java community and investing considerably in the continued development of Java. You could say the Open JDK has been the most successful open source project in history, you know, even more so than Linux.”
Solving the Java performance challenge was key
The move from hardware to software for Azul also tied in with the rise of cloud computing, Linux and commodity servers. For Sellers, that allowed Azul to bring its focus on performance and specific use cases to the way it built software. The impact of that, according to Sellers, is that now the company has all of the 10 top trading firms using its Java products over any other version of Java.
Financial services is a hard market. Performance is not just king, it is everything when trading companies are involved. What makes Azul different from its competitors or is this, like a lot of software, once installed, never removed?
Sellers accepts that need for speed. He points out that many of those financial services companies resort to using hardware such as FPGAs to meet their performance needs. But, he also sees trust from those companies in the ability of software to make a difference. He said, “That’s ultimately why Java has continued to be very, very popular and continues to evolve. A lot of applications rarely get rewritten, they get evolved.”
He continued, “Even in a very speed and feeds-driven market, like financial services, Java continues to have a very strong footprint. There are trading platforms out there that are not based on Java that are based on hard-coded C, C++ speeds and feeds using GPUs for the ultimate speed advantage. It’s a pretty narrow market segment, and one where we came from and have expanded considerably beyond that.”
How much of this is about pure app development?
It raises the question of exactly where Java is being used. The big database players like IBM and Oracle use it to front-end access to data. So is this pure app development?
Sellers replied, “It’s primarily app development. In that world where the data is stored is almost irrelevant, the persistency of the data is very asynchronous from the actual real-time operation of a trading platform. I wouldn’t characterise those markets as dominated by an Oracle database or IBM database or no SQL database or whatever. That’s because in that world, the data is always in flight and being held in memory as opposed to being persistent.”
How do you compete with Oracle?
Another one of the big bonuses for Java is that you can deploy any app into any Java Virtual Machine (JVM). Sellers said, “That’s one of the reasons that we are literally bug for bug compatible with Oracle. And that’s viewed as a good thing because no one’s complaining about Oracle’s software. It’s not bad software, it’s great software. It’s very robust, it works, its stable, otherwise, it wouldn’t be as successful as they have nor with Java in general.”
Competing with Oracle is a significant challenge, especially given Oracle’s marketing power and size. How do you take on a company that size and build a business as big as yours?
According to Sellers, “In the Oracle Java product, there is no feature performance management, there’s no differentiation in the product. What is in open JDK is what is in the Oracle product, and that’s true for us as well. In essence, we have a true like for like product with Oracle, which allows for correctly trivial migration, right for large enterprises.
“That is the trade-off. Oracle could offer a value-added product like we did. We have three main products. Platform Core as pure open JDK. It’s basically an open-source business practice for us. The code is an Open JDK, we build it, we provide it, we offer long term classic 24/7 support for our customers. That has been a great market for us.
“Platform Prime is our performance product. It’s also based on OpenJDK. But we enhance it. We have a faster JIT compiler, we solve the garbage collection problem, we have ways of starting and shutting down applications much quicker, these types of things. Could Oracle choose to do a value-added product like we’ve done with Platform Prime? Sure, but they haven’t. That’s something that they could consider at some point, but we certainly don’t see that happening.
“The third product we have is called Azul Vulnerability Detection, or AVD. It’s a SaaS product focused on application security.”
Oracle losing market share
The challenge for Azul is how to get customers to move from Oracle. The products described by Sellers show differentiation. The problem is, with Enterprise Software, companies are usually reluctant to change for fear of problems in production. Even when there is money to be saved, Sellers says that most will stay with Oracle rather than take a risk.
Despite that, things are changing. Sellers cites Gartner, who says that in 2018/19, Oracle had something like 60% to 70% market share in terms of the use of the Oracle Java runtime. This year, Sellers says, that is expected to be 30% and continuing to decline.
Azul has managed to capture part of that, but Sellers says that an increasing number of companies are prepared to move to free builds of Open JDK. He said, “A majority are just using free builds. You can get free builds from us, Amazon, and Microsoft Eclipse has a project called Tamurin. The fact that it is commonly available means that the majority of the world will just use free builds.
“Most of the industry analysts peg the paid Linux space as roughly 25% of the total market. That’s where we expect Java to settle out. There are always going to be enterprises that, regardless of free being available, still need a commercial support partner that provides stability.”
Java customers following the open-source trend
Sellers sees the Java market as complex. Customers are loathe to update versions if they don’t need to. That means it has customers that are running Java six and seven which means supporting versions of Java that were released in 2006.
This is where commercial support vendors come in. Customers are willing to pay for that long-term support to keep apps running. It also means that someone has to go back and do security updates for those older versions of software. This long term support model is one that has paid off for some Linux and other open-source vendors.
Customers are also keen to find a vendor that will support all those versions. They do not want multiple Java suppliers supporting different versions. It creates a management and logistical nightmare and creates security risks. Sellers says that leaves them with either Oracle or Azul.
He commented, “We actually are better than Oracle in that regard, in that they stopped support for Java six and seven. It represents less than 10% of the usage of Java, but it’s still an important 10%. Banks and retailers can’t afford to leave applications with security vulnerabilities. They need a partner, and Azul is the only vendor that offers continuing support for those older versions.”
To update or not to update
One of the challenges here is organisations identifying those older versions and where they are used. It is not a new problem, and that is where Software Asset Management (SAM) solutions come in. Sellers says knowing the versions is just the start of the issue. More importantly, customers need to decide what they are going to do about it.
For many customers, one question they need to ask is, how important is that application? Once they answer that, Sellers believes the next question is, upgrade or just maintain? Upgrading from very older versions of Java can be a lot of work, and that adds risk. It’s another reason why customers like companies such as Azul. They do not force the customer to upgrade, they just provide patches and support for legacy Java.
But when customers do want to update, Sellers believes that is, again, where Azul adds value. Its longevity means that it knows the problems with getting past the major upgrades to Java. He comments, “While Java is generally compatible, there was a fairly strong break after Java eight, where a number of features were deprecated. Updating from six to seven to eight is pretty straightforward.
“Trying to go from that to nine and beyond depends on the features that were used. There were some deprecated features, so sometimes it’s more work than enterprise customers want to take on. When our customers engage with us, they have access and support for whatever version of Java they want to use. If they want to upgrade, we let them make that decision.”
Overcoming organisational skill shortages
What is interesting is that despite the longevity of Java and how commonly it is taught, there are skill shortages emerging. Supporting the older versions with features that have been deprecated means having people who understand that code. Enterprises are busy training staff for the latest technologies in their stack. They don’t have time to train on older tech.
It creates an interesting problem for Java. Java is still evolving, but developer rates are falling because it is not the latest technology. Some developers will be happy updating their Java skills, but others will look at training that give them higher-paying skills.
The result is another opportunity for companies like Azul. Sellers said, “Java is moving faster than developers can consume it. Every six months, coming up with new features, new capabilities, etc. It’s moving really, really fast. It has reenergized the interest in Java and the use of Java.”
The result of this, according to Sellers, is that enterprises have the best of both worlds. Its success in the enterprise makes it hard to remove because no one wants to rewrite an application. But if developers are not keeping their skills up, where will enterprises get their development skills from.
As Sellers pointed out, Azul is second only to Oracle in the number of software engineers focused on Java. It gives customers not just the ability to buy long-term support for older versions of Java, but also hire in development skills.
Reducing costs is not just about cloud computing
The use of Java for business and transaction logic makes it ideal compared to other languages out there. But that means you need developers who understand the intricacies of Java. This is especially important when you are creating apps where performance is critical.
A lot of companies have been seduced into thinking that moving to the cloud and having access to seemingly unlimited resources will overcome performance issues.
While talking about recent releases, Sellers commented, “For customers that are deploying in the cloud, being able to have a faster JVM allows them to reduce the amount of compute needed in the cloud. It immediately saves on your Amazon bill. We see customers who are using Platform Prime to reduce their cloud builds of Amazon by 30, 40, and 50%.
“It’s a phenomenally wide use case for us. That part of our business is growing exceptionally well.”
Securing Java
Azul recently added its first SaaS product, Azul Vulnerability Detection (AVD). Sellers said, “We have all sorts of really interesting information inside the JVM in terms of what’s actually running. Our heritage is all about doing things in a very performance-centric manner. Customers were telling us about problems that they faced related to security vulnerabilities that may exist.”
With all the security tools that are in the market, it is easy to think that IT Security Teams and developers already have this covered. But for many, that isn’t the case. While the tools flag suspicious libraries in development, once something gets into production, it is often not scanned or checked as new vulnerabilities appear.
Sellers commented, “Since JVMs are running applications, we actually have precise information that we can provide to the customer in terms of exactly what they’re running and whether or not they’re running vulnerable code. Regardless of where the application came from, it could be an in-house developed app, a third-party app, or an app that was developed 20 years ago.
“So long as it’s running Java, we can communicate that information to our SaaS service, and map on a very precise level, literally down to the method level. Is this application running a piece of code that matches against a known vulnerability in this database? We provide that information in real-time to our customers.”
Creating a risk-based approach to patching
What makes this especially interesting is not that it identifies the vulnerability but that Azul is able to pinpoint where it is. More importantly, it also determines if you call the vulnerable piece of code. That allows an organisation to take a risk-based approach and decide when and if to patch it.
As Sellers points out, if you have 1,000 Java apps in your enterprise, suddenly trying to patch all of them is impractical. Just because the library you are using has a vulnerability does not mean you are using the piece of code that is the problem. If not, you do not need to patch. Organisations can now make better decisions that enhance security without disrupting the business.
Enterprise Times: What does it mean?
Over the last few years, Oracle has continued to tinker with and increase the cost of its Java licences. At the same time, the amount of Java-based code inside organisations continues to grow. By offering faster, optimised and more performance-orientated code with new security tooling, Azul is offering companies an alternative to Oracle.
As Oracle’s share of the market continues to shrink, it will be interesting to see how quickly Azul grows its share. The company is fast approaching a $1 billion valuation making it a tech unicorn. It is proof that for open-source, there is plenty of room to create an enterprise business.
