BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

FICO Proves The Mainstream OpenStack Adoption Point

This article is more than 8 years old.

FICO is a nearly 60-year-old Californian company. One doesn't expect to see 60-year-old companies provide the best case-studies for cutting-edge technology adoption but recently I took the opportunity to talk to FICO about its attempts to transform itself from an old-world model to a new one. I came away impressed by the company's vision and execution and got a good story about cloud adoption to boot.

FICO is best-known for creating the FICO Score, a measure of consumer credit risk. Founded by engineer William Fair and mathematician Earl Isaac, the company began selling credit scoring systems back in 1958. The two met while working at the Stanford Research Institute in Menlo Park. FICO went public back in 1986 and its use of Big Data and mathematical algorithms to predict consumer behavior has been applied to a range of different industries.

FICO’s software technology was, however, very traditional in nature. A couple of years ago FICO came to the realization that, since its value was really contained in the intellectual property encapsulated in its software, it needed to transition to a more services-based model. The company brought in Mike Trkay to help it transition its infrastructure model and team from a traditional on-premises model to a services-based one. The idea being that this would give FICO the flexibility to innovate quickly. FICO already dominated the Tier 1 financial services market, so it needed to expand from this saturation and go broader and deeper. FICO looked at its growth opportunities and rightly pointed out that it needed to lower the barrier to entry such that it could address a market opportunity lower down in the food chain.

Trkay explained that the company wants to convert its product portfolio to a platform. All the forms, rules engines and decision models it has developed could be put into a catalog to let customers build their own apps based on FICO’s analytics components. The company is one year into that process and demand is accelerating. Trkay indicated that this is a case of "skating to where the puck will be" and creating a platform that should be complete by the time customers are ready for it.

So to the technology. Historically FICO used VMWare as its only virtualization vendor. FICO identified that VMware would not be able to be its vendor of choice going forwards - FICO has a widely dispersed business and regulatory compliance required that points of presence were needed in-country. Building new presence in global territories would be an expansion of a required virtualization footprint.

FICO looked at the option of proprietary cloud solutions, but simply couldn't get the economics to work - it is, after all, looking to lower barriers to entry and hence pricing is critical. FICO quickly looked at OpenStack as its core infrastructure solution and began to assess vendors. FICO ran an interesting process whereby parallel teams would assess service-based and hyper-converged offerings.

Eventually FICO settled on a stack that included RedHat as its OpenStack vendor, Red Hat's OpenShift as its Platform as a Service (PaaS), with storage based on a dual approach - high-flying SolidFire for Tier 1 and Ceph for its Tier 2 and 3 needs. Studious readers will notice that, other than the SolidFire components, all of these pieces come from Red Hat. This raises two questions in my mind: how much was the technology decision-making process predicated on a single-vendor strategy and why did SolidFire make the cut as an external party?

On the first point, the answer is in the evolving design process used by FICO. The PaaS decision was made in advance of the OpenStack and storage decisions. PaaS options were very immature in 2013 and while FICO is still confident in its technology direction, the advancement around containers and container orchestration solutions in the last two years has been of great interest to the company and its use of OpenShift as part of its core platform. In particular, FICO is interested in how OpenShift v3 will incorporate Docker. After selecting OpenShift, FICO later made a decision on RedHat OpenStack based on the bake-off between services-based and hyper-converged models. Ceph (pre-Inktank acquisition by RedHat) was selected as part of that services-based design. SolidFire was the latest evolution of the design based on continuing performance testing of the platform against FICO requirements.

In terms of the SolidFire decision, the company was keen to use Ceph for all its storage needs but simply couldn't get the performance to where it needed to be - hence the move to SolidFire. SolidFire also delivers far more flexibility in terms of geographic growth for FICO (i.e. scale Tier 1 storage in each data center to align with varying in-country volumes). The FICO team even looked at running Ceph on solid state drives but found that to be an expensive approach without the de-dupe and compression capabilities available within a solution such as SolidFire. FICO needed performance for a transactional database. It went through a number of vendors but SolidFire was the only one that checked all the boxes for the company.

Interestingly FICO has a fairly traditional approach towards networking, choosing Nexus 9K. The team felt it gave them a solid network base with the potential for Software Defined Networking (SDN) in the future as SDN capabilities matured. Again, it would be interesting to see which direction the company would go were it making its decisions today; SDN has come along in leaps and bounds since then.

The final determinant on the success of this journey is a telling one. Trkay told me that FICO is now broadening this initiative from customer-facing requirements to corporate data center ones. The performance, costs, and flexibility gains that it has seen are attractive and FICO wants to move forward from its legacy corporate IT platforms from VMWare and NetApp. Trkay told me that he feels FICO has found a good balance between cost and performance.

Follow me on TwitterCheck out my website