It seems like everyone who ever sold a system or network management tool is now repositioning themselves as a controller of clouds. Gale Technologies, which you probably never heard of because it hasn't been marketing itself for very long, is coming to clouds from the network side and expanding out into controlling all physical and virtual resources, including provisioning of servers, storage, networking, operating systems, and applications.
So basically, we're done, right? You can all go home.
Well, probably not. For a while at least, people are still going to be an integral part of the tools that Gale Technologies and others are peddling.
Gale Technologies was formed in June 2008 from the assets of two network provisioning companies called Edentree Technologies and QuickCycle, both of which sold "lab management" and automation tools for network equipment. These lab management products are designed to allow for development and test environments to be raised and torn down quickly to speed up the creation of software or services. There are lab manager programs from Citrix Systems and VMware, which both companies acquired, to do similar jukeboxing and deployment of virtual machine images on a pool of physical servers.
With GaleForce, Gale Technologies is able to provision virtual or physical servers, their storage and networking, and just about any device you can put on a network. Garima Thockchom, vice president of marketing at the company, tells El Reg that it has created over 130 different resource adapters, which are agents for the GaleForce controller server that allow it to reach into very specific devices and bark orders at them or, in the case of servers and hypervisor partitions, shove operating systems and software stacks on them. Once the tool is set up and resource pools are created, you are talking about somewhere around four hours to provision a stack instead of one or two weeks using manual processes.
The GaleForce controller isn't just for clouds, since it works equally well for the underlying physical devices, and that is a key differentiator compared to a number of point products that are on the market and that are focused mostly on virtualized servers, storage, and networking. In many cases, these companies forget that there are plenty of Unix, mainframe, and other proprietary servers out there, grinding through work, that also need to be smacked around by what amounts to a traffic cop so they don't behave badly.
The list of resource adapters for the GaleForce controller is wide, but it is still not wide enough. It includes x64-based servers from Hewlett-Packard, Dell, IBM, and Oracle (well, they were created for Sun boxes when they were still Sun); storage arrays from NetApp and EMC; network switches from Cisco Systems, Juniper, HP, Dell, F5 Networks, Redback (this includes in-rack gear as well as L1 and L2 VLAN switches); storage area networking switches from Cisco and Brocade Communications; hypervisors from VMware and Citrix; power controllers for the data center from Avocent, APC, MRV. WTI, and others; and public clouds such as Amazon's EC2 and Rackspace Hosting's eponymous Cloud.
You'll notice that list does not include any RISC/Itanium Unix iron, IBM mainframes or AS/400s, and a slew of other networking and storage gear. But, Thockchom says that the techies in Gale Technologies' professional services team can build one for any device you have in your network in about two to three days, or if you are a DIYer, you can use the software GaleForce development kit and build it yourself.
The resource adapter list also does not include Hyper-V and KVM hypervisors, which are in the works, or Integrity VMs, Solaris containers and LDoms, PowerVM guests, and mainframe LPARs. The latter should be on the list if Gale Technologies doesn't want to just play in the x64 world.
The GaleForce controller is itself coded in Java its APIs are exposed in XML. System administrators and infrastructure architects can create pools of infrastructure using the GaleForce Control Center, create templates that describe stacks of software and hardware that are to be managed as a workload, design workflows that controls how these resources are provisioned and moved around the infrastructure pool, which can be a physical or virtual cloud or a mix of the two. The APIs available here are XML, but the company knows some people prefer Python (and some public clouds do, too), so a set of Python APIs is in the works for GaleForce.
GaleForce also has a Web portal service catalog that lets line of business managers and other designated employees in the company, such as those in control of test and dev, reserve and schedule resources and to automatically return bare capacity back to the pool when those jobs are done. At the moment, GaleForce cannot provision applications onto bare-metal servers, but it can provision apps on virtual machines.
Interestingly, when provisioning and cloudbursting out onto the Amazon EC2 cloud, GaleForce is not hooking into Amazon's Virtual Private Cloud, but instead uses Vyatta's open source VPN capability to create a tunnel to reach back into raw EC2 images.
Gale Technologies has over 100 customers today, with the majority of them using its Edentree and QuickCycle tools still. But in the last year, since it started selling the broader GaleForce controller, it has cadged 22 customers, and many of them are high profile customers - such as Cisco, NetApp, Juniper, SAS Institute, AT&T, BT, and Time Warner Cable - who are using its products to build up and rip down test and dev environments inside their own companies. They are not, as yet, using GaleForce in production IT environments. But remember, server virtualization started out in test and dev and took many years to move into the data center. You test it on the coders first, then they tell you if it is good enough for production.
Thockchom says that a number of companies are starting to use the GaleForce controller to take a snapshot of their production and backup environments and throwing the whole shebang out onto an Amazon EC2 slice to check whether their failover procedures for those production environments work. The reason they want to do this is because in a lot of cases, both the production and failover environment in the data center (or data centers) are actually both running workloads. To do such testing on them makes CIOs and CEOs very nervous.
This week, GaleForce added preconfigured templates to the stack, such as for Dell services using Cisco switches and NetApp storage or IBM servers using Juniper switches and EMC storage. There are sixteen of these preconfigured "turnkey clouds" (and yes, my dyslexia keeps making that turkey clouds) available now. With these templates, the company says it can turn current infrastructure in your data center all cloudy (by which it means into a pool of resources) in about two weeks.
The GaleForce 5.3 controller server and the turnkey cloud templates are available now. A package spanning 15 physical devices with licenses for up to 60 virtual machines costs $50,000, including two days of professional services for setting it up, training for admins and users, and one year of support. This server can span thousands of different resources on your network. ®