Original URL: https://www.theregister.com/2013/02/08/red_hat_legacy_maker/

JBoss is juicy, but Vert.x could bring sexy back to Red Hat

Linux shop man on app servers, business and keeping devs interested

By Gavin Clarke

Posted in Software, 8th February 2013 11:05 GMT

Seven years after Red Hat snatched JBoss out from under Larry Ellison’s nose, the enterprise Linux distributor is continuing to squeeze the juice from the open-source application server.

Red Hat spent $350m buying JBoss in 2006 and today it forms the technology backbone and the brand-name basis of Red Hat’s enterprise middleware suite.

And, as Red Hat grows, so does JBoss – with customers who pick Red Hat Enterprise Linux increasingly making JBoss their app server of choice.

Werner Knoblich, Red Hat’s EMEA general manager, told The Reg yesterday that revenue for the JBoss application server is growing at 40 to 50 per cent per year.

And for all the work that Red Hatters put into building and buying the JBoss Enterprise Middleware suite, it's the application server that remains the door-opener for customers and makes the lions' share of the money.

The bread-and-butter business is JBoss as a replacement for IBM’s WebSphere and Oracle’s WebLogic – particularly WebLogic. As Oracle’s chief executive gobbles up more technology, the customers using WebLogic are becoming increasingly eager to diversify the portion of their corporate IT wallet currently funding Ellison’s yachts - not least because the JBoss option is a little cheaper.

Business is particularly good in the wake of 2008, as IT types have lost their complacency over endless enterprise renewals with the contraction of budgets. Red Hat charges for JBoss on a subscription basis while IBM and Oracle charge for cores and then for maintenance. Red Hat reckons its subscription-based support for JBoss – it doesn’t charge a licence as it's open source – is up to 50 per cent cheaper than the maintenance prices of IBM and Oracle. And that’s excluding the charge IBM and Oracle level for licensing.

It’s a decent enough business just hauling in replacements for IBM and Oracle's kit: WebSphere and WebLogic were, between them, the number one and number two app servers of the day when Gartner still counted application server market share back in the 2000s.

JBoss has paid off big-time for Red Hat, Knoblich tells The Reg.

Then came the cloud and dried up all the rain

But its money-spinner is slowly becoming a problem for the Linux-slinger too, because while business is booming, it’s booming in market that’s not growing. Replacing WebSphere and WebLogic is a rich seam to mine because there are plenty of customers who want to run their own apps on their own servers, but it’s a finite seam. It’s the cloud that’s brought the change, as middleware becomes disintermediated by the presence of Amazon and OpenStack APIs, with their elasticity. If, and when, OpenStack explodes behind the corporate firewall, then JBoss will find itself hitting a wall on sales – the question is how long it’s got left.

Meanwhile, Red Hat’s drive into the enterprise has seen the JBoss app server lose the excitement and buzz that once surrounded it in the eyes of developers who'd helped it in the early 2000s and who saw it undercut IBM and BEA Systems – which gave up five years ago and was eaten by Oracle.

In those early days, JBoss wasn’t just a novelty but - used in conjunction with Hiberate - it was extremely useful and powerful: JBoss and the Hiberate framework gave devs a refreshing new way to build apps - cheaply and quickly - using open source and a with a modular framework. This contrasted to the expensive, monolithic Java Enterprise Edition frameworks pumped out by IBM, BEA and Oracle.

The combination of JBoss and Hibernate drew developer adoption and contributions to the projects, which in turn further stimulated interest and adoption.

Hiding it under their hat

What’s Red Hat got on that scale today? IBM-style solutions are Red Hat’s middleware mantra, targeting the Business Process Management (BPM) and Service-Oriented Architecture (SOA) consultants and experts.

The Linux shop does have something – it’s called Vert.x, but there’s a problem.

Vert.x is hailed by some as the next evolution of the application sever; it’s a modular and multi-language framework that runs on a Java Virtual Machine (JVM). While some are making the JVM multilingual, Vert.x goes further: it lets you mix and match programming languages in a single application, so you can code in Java, JavaScript, Groovy, Ruby and Python.

Vert.x could also eclipse the industry’s most recent best hope, Node.js - which allows JavaScript to be used end to end, both on the server and on the client. But’s server-side Javascript can also be a real challenge when it comes to scaling. The polyglot Vert.x, meanwhile, is intriguing devs, and its ranking rise in Google searches reflects the growing interest.

Vert.x is an interesting and potentially disruptive technology and it’s generating interest among those who 10 years ago would have been championing JBoss.

Red Hat has recognised this and in traditional JBoss form (dating from when JBoss hired the core maintainers under chief Mark Fleury), Red Hat recently hired Vert.x creator Tim Fox, who quit VMware in late 2012.

And in a demonstration of just how serious Red Hat is about Vert.x, the Linux company – which is no pushover when sabres start rattling - stood up to VMware, which had claimed that it owned Vert.x because Fox had developed it while in VMware's employ. Red Hat and VMware have now indicated that they will now work out the best way to manage the Vert.x community going forward.

That’s all peachy, but what does this mean to JBoss?

In the process of replacing IBM and Oracle, Red Hat has now become the new legacy. The question will now become how far and fast it pushes Vert.x and if it continues development of JBoss. Should it have two app servers - and support the old JBoss while pushing Vert.x - or bring JBoss to an end?

The latter will jar with the kind of enterprise types Red Hat reckons says pay for the freebie application server to ensure support and reliability.

But pushing Vert.x could help re-invigorate Red Hat’s pull among early-adopter devs who - back in the day - helped make JBoss and disrupt the competition. These are the same developers which enterprise types now pay Red Hat to support.

It’s a balancing act Red Hat will need either to perfect or abandon. As Knoblich, who has been with Red Hat since 2003 in the pre-JBoss days, told El Reg of Red Hat’s attempts to remain both sexy and serious on JBoss: “We have to be very attractive to developers but we also have to monetise things and we have to sell or enterprise products to IT departments,” he said. ®