Interview with David Bernstein
David Bernstein sheds light on Cloud Computing and its role in the development of the Smart Grid.
Question: Why should utility executives care about Cloud Computing?
Cloud Computing is an ideal complement to the Smart Grid. It would appear to me that without cloud operating a fully realized Smart Grid using traditional IT architectures and paradigms would at the very least be expensive — and perhaps even impossible.
Question: Why is that?
The torrent of data and the wide swings in demand for computing and storage resources would require big IT investments if every utility decided to go it alone. Unfortunately, the equipment purchased or leased by each utility would probably have a fairly low utilization factor. Even if a very substantial investment were made, an unanticipated spike in demand — something utilities know a lot about — could compromise the whole system. Because the Smart Grid requires interaction between utilities and both private and public entities, the applications used by all parties use should be consistent. Revisions and bug fixes should be applied only once and be universally available immediately. The challenge for version control for Smart Grid applications in a traditional IT setting seems to be quite onerous.
Question: You mentioned traditional IT architecture and paradigms more than once? What is cloud computing? How does it differ from our current IT operations?
You can think of Cloud Computing as having four characteristics that make it distinctly different from conventional IT as: Automation, Virtualization, Redundancy and Autonomy. There are other differences but those four seem to me to the most demonstrative.
Question: Let's do them one at a time. What do you mean by Atomization?
Think of a vast collection of computing, storage and networking components (including software operating systems and applications) that instead of being integrated into a unitary system are considered as individual elements. From this perspective, computing resources have been atomized. Cloud software configures these elements to meet the instantaneous IT requirement of each client with their very own virtual data center.
Question: You mentioned Virtualization. Could you expand on that concept?
At any given moment your application may be in Cleveland, your storage in Baltimore and your CPU in Phoenix. Cloud software keeps track of it all and reconfigures your virtual IT requirements change. In utility parlance, this is somewhat analogous to load optimization.
Question: Atomization and virtualization would seem to make Redundancy critically important.
Correct. While redundancy is often present in conventional systems, Cloud Computing takes it to a new level. All of your data and virtual IT configuration information – along with everybody else's – is stored in at least three different places. The software that runs the Cloud knows where it is and will find it if one or more of the atomized computing resources fails.
Question: What does all this have to do with Autonomy?
The cloud's normal operating mode is auto pilot. When a computing element dies there is no attempt to fix it. Instead, the software that runs the Cloud simply ignores it and allocates its function elsewhere. Experience has shown that reliability and availability metrics are most adversely affected by human error (connecting the wrong wires, forgetting to install a software patch).
Question: Are there any special implications for utilities underlying these characteristics? If so, what are they?
Utilities operate under a regulatory umbrella that varies from state to state. The first issue to spring to mind is privacy. Here are some questions utility executives have to ask: What does the PUC allow regarding collocation and co-tenancy of data? Does the PUC understand Cloud Computing and, more specifically, its potential value to utilities and ratepayers? What privacy guarantees will be required to allow utilities to use cloud?
As Cloud Computing becomes more accepted and its ability to protect data more nuanced, regulatory bodies may well address privacy issues in a new context that takes Cloud Computing’s advantages into consideration. This should not be viewed as loosening privacy restrictions but as acknowledging that the Cloud's "automatic anonymity" can provide powerful privacy solutions.
In the meantime, some utilities are exploring private clouds — virtual data centers for and by utilities with the same privacy and accountability issues. Within a private, utilities-only cloud, data can be secured at appropriate levels. Individual utilities or departments within utilities can specify the sensitivity of their data and allow the Cloud to process and store it in the most cost-effective manner according to defined rules based on parameters such as location of data centers and authentication.
Question: Cloud Computing is still new. What are its greatest challenges to its success and acceptance?
Like any important leap forward in technology, the promise is made before it can be fulfilled — and Cloud Computing is no different. Cybersecurity should be a primary concern of any utility executive. Cybersecurity includes data loss and data corruption either through an attack, human error or severe environmental events such as earthquakes and hurricanes.
Compromised data poses a different type of problem because it will often involve human intervention that can be intelligent, subtle and effective. The Cloud's built-in redundancy addresses these challenges directly and probably makes the Cloud a safer place to store data than an individual utility can manage with backup drives and other traditional methods. While it is natural to focus on the Cloud's characteristics — such as multi-tenancy of data and software — as potential problems, the Cloud can just as easily provide powerful solutions to cybersecurity challenges. Substantial hardware, software and human resources can be brought to bear within the Cloud that are far beyond the resources a public utility could hope to deploy on its own.
Finally, like any data center, Cloud-based solutions can become overloaded, which will degrade performance. From a utility executive's perspective, performance issues can best be addressed through two strategies: Contracts and exit plans. Including carefully considered incentives and disincentives in contracts with Cloud providers will help maintain adequate performance. In the event that contractual guarantees are consistently violated, a well planned exit strategy should be in place.
As VP of Strategy for Cloudscaling David works on Corporate Strategy, Key Customer Engagements, Standards, and Government Affairs.