Makara's do-it-yourself platform cloud
Like Google App Engine. Without the Google
So, you want to use so-called platform cloud like Microsoft Azure, Google App Engine, or Engine Yard? But you want more choices and more control than these application clouds offer, and you don't want to have to monkey around and build a virtual data center yourself using Amazon's EC2 or a number of infrastructure clouds? A startup called Makara is cooking up something that you might find interesting.
The Cloud Application Platform, which is launched today as a beta program by Makara and which goes by the alternative name of WebappVM, is yet another abstraction layer that rides on top of iron to make it easier to use and manage. The idea, says Isaac Roth, co-founder and chief executive officer at Makara, is to allow PHP and Java applications that have been coded to run on systems inside companies to be deployed without having to change their programming models, as application cloud providers do.
The WebappVM abstraction layer replaces operating systems that would normally run atop virtual machine hypervisors, operating systems that in turn support applications. WebappVM includes PHP and Java runtime environments as well as a middleware stack based on either Tomcat or JBoss application servers; Apache is the Web server of choice.
This abstraction layer hooks into the provisioning APIs used by public cloud providers or to the provisioning features of the popular hypervisors used on commodity x64 servers that are generally used to build private, internal clouds. Makara has deployed whatever technique necessary to talk to public cloud and hypervisor provisioning tools, be it C, Python, or Web services methods.
The WebappVM abstraction layer also has features to gather performance information on running applications and automatically scale resources on the clouds as PHP and Java applications run, and it also has change management features that allows for upgrading of the runtime and middleware environments and rollback when something goes wrong. The WebappVM tool has a user interface coded using Adobe's Flex tool for creating Flash applications. The upshot is a tool that can manage different and incompatible infrastructure clouds and hide their differences to PHP and Java applications.
The WebappVM is currently supported on the Amazon EC2, Rackspace Cloud, Terramark vCloud public clouds and can be used to create an platform cloud internally on VMware ESX Server, Oracle VirtualBox (formerly Sun), or Xen hypervisor (it can be Oracle VM, XenServer from Citrix Systems, or the Xen hypervisors embedded in Linux by Red Hat and Novell). The WebappVM extraction layer does not, as yet, present a database service layer, but the applications that are running atop WebappVM can be pointed at any database source they need on the public or private cloud.
Roth, who spent five years at application performance monitoring tool maker Wily Technology before CA ate it for $375m in 2006, co-founded Makara with Tobias Kunze Briseño, the company's chief technology officer who used to run portal development for Lycos Europe and who used to run software engineering for digital media at supercomputer maker Silicon Graphics. Roth has worked as a software engineer at both Cisco Systems and Red Hat in their early days.
The company has venture backing from Shasta Ventures and Sierra Ventures as well as the Marc Andreessen-Ben Horowitz tag team. Makara was founded in May 2008, originally as OSS-1701 (yes, that's the Open Source Software flagship Enterprise) and then briefly by its product name, WebappVM.
Roth says that he wants beta testers to kick the tool around for a few months, and once it is hardened Makara will release WebAppVM 1.0 for production use. Pricing has not been set yet, but on public clouds, Makara will charge an incremental metered usage fee on top of the cloud provider's infrastructure fees, and on private clouds, Makara will charge a fee for every virtual machine in production. You can download Makara's Cloud Application Platform here and see a YouTube video walking you through the features there. ®
Sponsored: Hyper-scale data management