Original URL: http://www.theregister.co.uk/2010/12/09/salesforce_why_buy_heroku/

Salesforce wants some Ruby love

$212m goes a long way

By Gavin Clarke

Posted in Developer, 9th December 2010 06:25 GMT

Dreamforce 2010 Why would a company that spent 10 years delivering CRM as a service drop $212m in cash and $27m in stock on a 30-person startup that hosts Ruby on Rails apps for devs on Amazon?

Salesforce.com vice president of technology and co-founder Parker Harris was initially, oddly vague on Wednesday when explaining why his company just bought three-year-old Heroku.

During a Q&A with press and analysts at Dreamforce in San Francisco, California, he was asked on why Salesforce.com is lavishing so much on so few. "We want to hear what our customers want us to do," he said.

But you don't just drop more than $200m on a company to see what sticks. We're sure Harris has a pretty good idea what he's after.

The deal is certainly odd. According to Harris, nothing's going to change on the Heroku end. The company, which serves more than 105,000 Ruby apps on top of Amazon's compute and storage infrastructure, is staying on Amazon and not shifting to Force.com.

"It's wrong to think moving Heroku to Force.com or Force.com to Heroku. We've expanded. You have to think of this is an expansion of what we are doing. In some cases it [apps] will be built on Heroku, VMforce, App Builder," Harris said.

And Heroku will be given plenty of scope to do its own thing. It will retain is current offices in San Francisco, and its existing roadmap, Harris said.

"We aren't taking these guys and saying come in the building and we are going to rework your roadmap and we will tell all our customers the roadmap you are getting," Harris said. "These guys already have a roadmap - it's fantastic."

So far, the company has aimed to partner with those who bring technology add-ins to the basic Heroku service and to provide developers with additional capabilities. These include a Memecache add-in from Membase, formerly NorthScale, in April. You can see more here.

So far, so fluffy. But back to the question. What does Salesforce.com get for $212m plus $27m in stock?

In short, Salesforce is buying its way into developers and those who know them. Salesforce.com's background is sales people using its CRM software-as-a-service - still it's most successful service - and other line-of-business apps. The closest the company has come to developers is providing suits with some tech skills a drag-and-drop development environment they can use to customize their Salesforce.com or Force.com environments.

Real developers are coding apps, not assembling them using Salesforce.com's proprietary and closed and "Java-like" Apex programming language. And Salesforce.com knows it.

"Look at this conference - this conference is a lot of business customers," Parker said. "That's what this transaction is about is giving that value to developers. They [Heroku] get it. They speak to developers. We have yet to have that right voice, right technology and that right mindset.

"In terms of value, it's about a long-term strategic play. We believe it will be part of a huge platform strategy for us in the long term," Harris said.

Turns out Salesforce.com does know what it's after.

Real dev love

Salesforce.com has embarked on a strategy of turning the database and computer power underneath its core CRM service into something that can serve "real" devs.

This week, it announced Database.com, a branded version of its underlying relational, multi-tenant data store with its own URL and pricing. It also announced VMforce has entered private beta testing with release due in 2011. VMforce is a planned piece of technology, announced earlier this year and started by VMware's SpringSource division, to deploy Java apps - not "Java-like" apps - on Force.com using a virtualization layer.

Salesforce.com reckons VMforce has opened its eyes to a world of more developers and more languages.

In short, Salesforce.com is now doing what ISVs in the non-cloud world have tried to achieve countless times in the past: to build out a ecosystem of partners feeding off of its platform. Salesforce.com wishes to secure its place in the enterprise by achieving two goals: bringing more developers who are looking for customers on to its platform and attract more paying customers onto the platform in search of new apps as services. In previous years at other companies, the platform in question has been Windows, Linux, development environments, application servers, and databases.

Heroku's chief executive is Byron Sebastian, who in a previous role developed a web-services tooling environment for Java at BEA Systems. BEA's leadership embarked on exactly the same odyssey of trying to build out a partner ecosystem, but they failed and BEA's now in the possession of Oracle.

Salesforce.com believes Heroku can rub off on it and help it figure out how to become more of a Ruby service provider. Salesforce.com is therefore buying the brains and the technology of Heroku as it believes Ruby is poised for stellar growth as the language of the web and online services.

Heroku itself has grown to 105,000 apps from 50,000 in the spring, and boasts US electronics outlet BestBuy among its flagship customers.

"The big thing you get is you will trust the two [Salesforce.com and Heroku] together more and more and we will find ways to deliver that rust to you," Harris said.

George Hu, vice president platform and marketing, added: "We expect them [Heroku] to look at our assets, and customer relationships, and tell us what they think is the best way to leverage that."

Harris, the man who built out Salesforce.com and Apex, seems genuinely awed by the Heroku platform.

Heroku is a multi-tenant hosting environment that runs Ruby apps on multiple servers using something that the company calls a Dyno Grid. The Grid uses POSIX, a Ruby VM, Mongrel app server, web-server interface, and an optional middleware layer, and it runs with different Ruby frameworks, like Rails and Sinatra. A Dyno is a single process that runs your Ruby code on the underlying server grid. Heroku sits on top of Amazon, and the big sell is its simplicity on features and price: you provision Dynos across different server CPUs rather than provisioning and working out the price for individual servers, as you would if you were using Amazon.

Harris said Salesforce.com wants to take its own tooling and break it down so there's lot of "small, sharp tools that are right for the job." It also sounds like there might be more tools serving application lifecycle management, security, and integration planned, but that these would come in the form of more Heroku-style add-ons. Harris ruled out any Salesforce.com-branded HR or ERP applications built using Ruby.

Salesforce.com's EVP of technology encouraged everybody to think of Salesforce.com as a plug in to Heroku. "This has gone very quickly," Harris said of the Heroku deal while looking ahead to what's next. "We now have to rationalize the development lifecycle. I don't have a great answer for that today and it's something I'm going to be working on."

When it comes to business application companies, developers are usually second-class citizens. The companies' target concern are sales people and CFOs using its apps, not those coding. Developers get a look in if the company's got a middleware strategy.

But Salesforce.com has realized it must be different and cater to real coders building real apps not business types doing drag and drop inside some templates if it's to grow and remain relevant in the enterprise.

Salesforce.com just spent $212m on a strategic bet. ®