IEEE Talks IoT: Chris Miyachi
Christine “Chris” Miyachi chairs the IEEE Cloud Computing Community. She works as a systems engineer and software architect at Xerox Corporation, and is a graduate of the Massachusetts Institute of Technology (MIT). In this Q&A, Miyachi discusses big data, the Internet of Things (IoT) and cloud-related issues.
Question: As the Internet of Things (IoT) connects innumerable data-generating devices across the world, do you hear concerns in the cloud computing community that current approaches and technologies have limitations?
Chris Miyachi: I do hear concerns and I think there are limitations. Cloud computing today in some ways makes it appear as if we have an infinite amount of data storage available. But people in the cloud computing field that I speak with say that, eventually, we may “fall off a cliff.” And they suspect that, possibly, IoT represents that cliff. I’m not an expert on IoT by any means, but the projections for big data in an age of IoT that I’ve seen certainly suggest that we’ll need to adapt existing strategies and technologies and develop new ones to meet that challenge.
One potential solution is to process data at the edge of the network, at or near the sensor level, and only return a portion of it for central storage and processing. Another strategy might divide up the data into manageable portions so that it’s logically partitioned (LPAR). Dividing a problem into smaller micro-portions will decrease the size of the data needed to draw meaningful insights and would be less taxing on the storage and processing side. These are known strategies that could be adapted to handle the big data generated by IoT.
Another concept that may come into play is inter-cloud computing, which belongs in that cut-the-problem-down-to-size category. If a single problem exhausted the physical limitations of a single cloud service and its infrastructure, separate clouds might handle parts of the problem, reducing the challenge to storage and processing. But a lot of issues in inter-cloud computing remain to be worked out, including interoperability and security, among others. Obviously we’re talking at a very high level and the necessary strategies and technical means to meet the big data challenge now or in an IoT-enabled world become very complex. I’d like to point readers to IEEE’s Cloud Computing magazine for more in-depth treatment of this topic. It has and continues to publish numerous insightful articles on this very issue.
Question: You’ve referred to storage and processing as possible limitations in an IoT-enabled world, which has yet to be fully realized. What are your current concerns with the cloud in this regard?
Miyachi: We do have an issue with the processing and analysis of the vast amounts of unstructured data generated by various embedded systems and currently in storage. Embedded systems have dedicated functions that are integrated within larger systems and they have limited abilities for processing that data. The data generated is often from legacy systems without the thought of analyzing it programmatically. These systems generate huge amounts of data that isn’t immediately used or processed. But there’s value to be gained by analyzing that data because it may contain insights of use to the respective enterprises that generate it.
I’d expect that unstructured data to grow in volume and, possibly, value, in an IoT scenario as well. So it’s a challenge now and going forward.
Question: On a practical level, cloud computing issues, such as data storage and processing capacity, appear to anticipate challenges that lay ahead with IoT. What other issues in cloud computing keep you up at night?
Miyachi: One issue is the potential for data in the cloud simply being lost or deleted. Another issue is the lack of transparency around how user-generated data is collected and applied by search engines and advertisers, among others.
Those are discrete issues, but they may be related in that they might be addressed by clearer, more transparent terms of service by the parties involved. I’m not sure that users of the Internet and cloud services understand that their data could conceivably vanish and that redundancy is critical, especially for enterprises. And I’d argue that many if not most of us have no idea how data on our Internet and cloud usage patterns is collected and how it is applied.
Whether the solution lies in regulation or market forces is a valid question. If and when the end-user is sufficiently concerned about these issues that they impact the bottom line of Internet and/or cloud-based enterprises, those enterprises might make transparent data policies a commercial differentiator. If you had a choice to do business with firms that make their practices transparent, versus those that don’t, I’d suspect that many of us would choose the former over the latter.