Bracket Computing bridges local servers, cloud capacity

Its approach isolates applications and their associated data so they can 'float' across multiple locations.


Stealthy startup Bracket Computing made its debut last month with more than $85 million in backing for its radical virtualization technology, which allows businesses to move computing workloads from cloud to cloud more easily.

The vision is to help organizations process applications wherever it makes the most sense from a cost or resource perspectives. The approach could also be used to bridge operations between on-site servers and those hosted in a cloud service.

"When you have a million servers at your disposal, you have the appearance of an infinite resource," said Bracket CEO Tom Gillis. "Our question to ourselves was: can we build a layer that could present applications with highly reliable and secure performance?"

Gillis was part of the founding team at security firm Ironport Systems (acquired for $830 million by Cisco Systems in 2007). Founded in 2011, Bracket's backers include strategic corporate investors GE and Qualcomm, as well as venture capital firms Andreessen Horowitz, Norwest Venture Partners, Sutter Hill Ventures, ARTIS Ventures and Allegis Capital. 

Bracket's technology, called the Computing Cell, isolates applications along with their associated data and other services needed to run them. Each cell can "float" across multiple cloud services. Bracket orchestrates the contracts with cloud service providers: the technology currently supports Google Compute Engine and Amazon Web Services, but other "hyperscale" services will be supported in the future.

Bracket has been working with a number of beta customers, many of them in the Fortune 500. One early adopter is FuseFX, a visual effects firm with offices in New York, Vancouver and the Los Angeles area.

FuseFX CEO David Altenau said his company has researched the notion of using clouds services for additional rendering capacity for years but the technical challenges involved were overwhelming.

"Each job we do is custom. The data sets are massive with custom fields and textures, and various set-up processes that are different," Altenau said. "There are customized files, for example, that might relate to unique textures or lighting."

Any given job might require 10 to 30 compute nodes. Using Bracket, his firm was able to create a server "footprint" in the Amazon cloud that mimics how the servers run locally. Whenever a job requires extra rendering capacity, FuseFX can call upon the cloud capacity. Each job can be controlled via an administrator desktop. The loads are encrypted, addressing security concerns; and they can be run simultaneously, so that work is completed more quickly.

Aside from avoiding costly server investments, FuseFX believes Bracket's technology will help its newer remote locations in New York and Los Angeles scale business more quickly. Space and power resources are constrained in those sites, so FuseFX must rely on cloud capacity to grow. "It's an absolute necessity," Altenau said.