Dean founded IC Manage in 2003, a company that he has helped expand to become the IC/SoC design and IP management technology leader. Dean was also founder, President and CEO of Barracuda Networks from 2003 to 2012, where he built and expanded Barracuda from a spam and virus firewall provider to a broad line enterprise technology company with more than 150,000 customers.
Dean was the founder of Boldfish, a leading provider of enterprise messaging solutions that was acquired by Siebel Systems in 2003. Dean was also founder, President and CEO of Design Acceleration, Inc (DAI), a maker of superior design analysis and verification tools. Cadence Design Systems acquired DAI in 1998. Dean was VP of Product Engineering at the 3DO Company and was instrumental in the development of the PowerPC architecture at Apple Computer. Dean received his BSEE from the University of Michigan, Ann Arbor and MSEE from the University of California, Berkeley.
We see the recent publication of the Blueprint for EDA Infrastructure white paper – how has the availability of on demand cloud computing changed the requirements of design management?
We are seeing very strong interest in hybrid cloud compute for IC design teams to keep their capital expenditures under control and be more agile in terms of bringing compute resources online quickly. Capacity planning is a hard problem with having to balance budgets, lead times and project schedule pressure. The widespread availability of commodity cloud CPUs provide a lot of promise, but EDA applications require a large amount of storage (tool trees, foundry PDKs, extraction data, etc.) and demand fast, shared storage as well. To get large data sets into the cloud takes days or even weeks and shared storage in the cloud tends to be both expensive and slower than on premise storage.
Design management for cloud must streamline the process of getting both repository and generated data to the cloud, minimize the cost of storing it there and use the cheapest, fastest possible storage devices that don’t create bottlenecks for 100s to 1000s of concurrent jobs. This is no small order and we are seeing great results so far with PeerCache for running a wide range of EDA applications in the cloud.
What have successful teams done to utilize the cloud to win?
Winning means reducing design costs, increasing efficiency of the design team, and getting more product to market faster. It’s not as simple as just picking a cloud vendor and setting up a VPN to run jobs. Costs can quickly spiral out of control in a cloud compute infrastructure and performance issues abound. The successful teams have leadership with a clear plan to leverage the cloud for either all or some of their compute needs.
Most of the successful large scale teams utilize their fixed cost, fixed capacity and fixed throughput on-premise compute infrastructure and augment this as needed with additional cloud compute.
They need to decide what types of jobs will run in the cloud. They need to establish a dynamic means of routing the jobs as appropriate. They have made the use of the cloud completely seamless for their engineers. Engineers don’t even necessarily know if the job is running on premise or in the cloud.
Having the data necessary to run a job in the cloud is the challenge. The winners have a central repository of design data and utilize IC Manage Peercache to make the managed design data and the non-managed, tool generated data appear in the cloud instantly for the job to run. They need the jobs to run without modification.
I’m happy to say that the investments IC Manage has made in our Global Design Platform, PeerCache and Envision products enable these teams to successfully leverage the cloud for competitive advantage.
What issues do globally distributed teams face for their design management and how do you solve them?
We’ve seen data management methodologies break down when teams tried to transition from geographically isolated design projects to a center of excellence or IP based design methodology. When the teams work in isolation, managing each project’s data in a local repository is effective, as long as they don’t need to send real-time or even daily updates to remote sites. When shifting to more agile methods to align available engineering expertise to meet project demands on a global scale, this localized methodology fails since it has dozens of repositories with conflicting versions of the same data and no clear genealogy.
As we pointed out in the IC Manage Blueprint for EDA Infrastructure white paper, the amount of data required to design, verify and tape out state of the art chips is just too large to manage in bulk. It’s 10s of millions of files, 100’s of terabytes of data that all have to be up to date, correct and consistent. Dealing with that in bulk is just impossible. We use meta data to increase efficiency.
Let me explain – using metadata allows us to organize our design components, their interdependencies and current state of completion in a high level configuration or component database. Rather that having to analyze millions of files to extract information, the newest version of our Global Design Platform called GDPXL provides high level web and automation frameworks that can query millions of relevant metadata records in seconds to answer questions like “who last edited the most recently released USB3.0 IP targeted for a specific 20nm technology.
What projects have used it so far?”
We represent very large devices in the 7nm and 5nm range and create data sets for design, verification, integration and tapeout at any site in a few seconds while maintaining 100% traceability of design changes. This is really a game changer for engineers, managers and executives to get accurate information and collaborate across the globe.
To deal with the sheer volume of both design repository data and generated data, our PeerCache product leverages metadata techniques to virtually project datasets to any remote site or into the cloud instantly. A metadata representation of the data that a user needs – again, EDA tools trees, PDK, design source, extracted results – is available in a few seconds and the great thing is that no expensive network storage is needed for that user’s data. This is how we help design teams have jobs up and running in public clouds in a few seconds instead of hours.
What issues do teams face in M&A and how do you merge design methodology?
We’ve certainly seen a large number of our customers in M&A scenarios over the past years. The biggest challenges happen when teams with completely different philosophies and methodologies are forced together. A lot of friction can be generated between teams that have invested deeply in automation and scalable data management methodologies as a strategic advantage vs. teams that have considered data management to be an overhead cost and live with inefficient design practices which add costs in terms of manual intervention and poor product quality.
The successful M&A projects we have seen consolidate onto a single global design management system quickly. They get their engineers working together and sharing IP right away on a single platform. They have clear leadership and a plan from the top. When our customers need it, we bring in a team of experts to assist or even integrate the data from disparate systems onto the IC Manage Global Design Platform. Our customers have
added 1000’s of engineers to their teams via acquisition and they have migrated them onto the IC Manage platform for maximum productivity and reliability in only a few months.
Platform flexibility, scalability, and expertise are key to making this happen so quickly. This is especially true in larger scale acquisitions.
How have you built IC manage to address these challenges?
I think the key thing to remember about IC Manage is that we were founded by hardware engineers that were both experienced and frustrated by the available tools for design management. Shiv Sikand our EVP of Engineering frequently refers to “People Who Care” as the ones that can drive change and produce great solutions and deliver great support.
After 17 years in the Design Management business, we have the industry’s most robust track record and strongest reputation for Design Management. We identify the “people who care” and bring them onto the team at IC Manage.
We pride ourselves on a level of service that is unparalleled in the industry. Our teams are available 24×7 with senior members to solve any issues our customers encounter in their global infrastructure. We stay ahead of the curve for our customers because they count on us to help them adapt through technology and organizational change.