Recently in San Francisco, Google had their Cloud Platform’s user conference which allowed attendees to hear from Joe Kava, VP of data center operations. Below are Joe Kava’s answers directly on how Google designs, builds, operates and secures their data centers globally.
Q: What is Google’s security model for data protection?
A: Google’s data center floor is equipped with laser beam intrusion detection, which is monitored 24/7 by high-resolution interior and exterior camera. If an incident occurs, activity records and footage is available.
Our security guards have also undergone rigorous background checks. Access to the data centers is only possible via a security corridor which implements multi-factor access and control using security badges and biometrics. Less than 1% of Google employees have access to our data centers. Our physical security features custom-designed electronic access cards, alarms, vehicle access barriers, perimeter fencing and metal detectors.
We track everything. From the time a HD goes into a machine until it’s verified as clean/erased or destroyed. Information security and physical security go hand-in-hand. As data travels across the internet or within networks it becomes vulnerable to unauthorized access. This is why securing data in transit is a high priority for Google.
Q: Do you build your own hardware and monitoring systems?
A: Yes, our production servers run a custom-designed operating system based on a stripped-down version of Linus. Google’s servers and their OS are designed for the sole purpose of providing Google services. Since our resources are dynamically allocated, this allows for growth and the ability to adapt quickly when needing to reallocate our resources for customer demand.
To aid our teams, we’ve built monitoring and control systems from our servers, storage and networking systems to our electrical distribution, mechanical cooling systems and security systems. In other words, we’re monitoring all aspects of performance and operations.
Q: How do you optimize your data center operations?
A: We are using our machine learning and deep algorithms for data center operations. In order to deliver optimal performance, our data centers are large and complex with electrical, mechanical and control systems all working together. It is impossible for us to visualize how to best optimize the data center in real time because of the sheer number of interactions and possible setting for these systems. However, it’s fairly trivial for computers to sort through possible scenarios for optimal settings.
We now also have an algorithm that is trained with billions of data points sites all over the world. Our machine learning models helps us visualize data and operations to be set up for the most efficient performance on any given day.
Q: How are you powering your infrastructure?
A: We are the world’s largest private investor in renewable energy. We have invested more than 2 billion in renewable energy Power Purchase Agreements. PPA’s are important because wind solar farms are on the same power grids as our data centers. Data centers sharing power grids gives the project developer the financial commitment they need to get the project built. Our investment is adding renewable power to the grid.
We also redesign our cooling technology every 12-18 months. Our innovations have developed water-based cooling systems such as seawater cooling, rainwater harvesting and thermal energy storage. We have other data centers that don’t use water-based solutions, instead they use outside air cooling. Each data center is designed for the highest performance and efficiency for their specific locations.
Q: Who operates your data centers?
A: Our employees manage and operate our data centers. We also take a different approach to the people we hire and how they run our data centers. They all come from very diverse backgrounds in engineering and operations. Many of our team members have experience in mission critical environments, this way they understand how systems interact together.
We also have integrated teams that are responsible for building capacity, commissioning the systems and providing 24/1 operations.