Last week my team hosted an exciting event at the Four Seasons Hotel in Houston, Texas. It was devoted to the continuation of the trend of developing closer relations between the participants. It was an event that brought together users, partners and customers. In addition, the event was attended by many representatives of Hitachi. When organizing this enterprise, we set ourselves two goals:
- Stir up interest in ongoing research into new industry issues;
- Check the areas we are already working on and developing, as well as adjusting them based on user feedback.
and Matt Hall) began by discussing the state of the industry and the various challenges associated with managing and processing seismic data. It was quite inspiring and, of course, significant to hear how the volume of investments is distributed between extraction, transportation and processing. More recently, the lion's share of investment has been in mining, which was once the king in terms of the volume of funds consumed, but gradually investments are moving into processing and transportation. Matt talked about his fascination with literally observing the Earth's geologic development using seismic data.
All in all, I think our event can be seen as a "first outing" for the work we started a few years ago. We will continue to inform you about the various achievements and successes in our work in this direction. Then, inspired by one of Matt Hall's talks, we held a series of sessions that resulted in a very valuable exchange of experience.
Edge (edge) or cloud computing?
In one of the sessions, Doug and Ravi (Hitachi Research in Santa Clara) had a discussion about how to move some of the analytics to the edge computing for faster and more accurate decision making. There are many reasons for this, and I think the three most significant are tight data channels, large volumes of data (both in terms of arrival rate, volume, and variety), and tight decision timelines. Although some processes (especially geological ones) can take weeks, months or years to complete, there are still many cases in this industry where urgency is of the utmost importance. In this case, the inability to access the centralized cloud can be disastrous! In particular, issues related to HSE (health, safety and environment), as well as issues related to oil and gas production, require quick analysis and decision making. Perhaps the best way is to show this with various numbers - let the specific details remain anonymous in order to "protect the innocent".
- Last mile wireless networks are being upgraded in places like the Permian Basin, moving channels from satellite (where speeds are measured in kbps) to 10 Mbps channels using 4G/LTE or unlicensed bands. Even these upgraded networks may not be able to handle terabytes and petabytes of data at the edge.
- Sensor systems from companies such as FOTECH, which are paired with a host of other new and legacy sensor platforms, are capable of producing several terabytes per day. Additional digital cameras that are installed for security surveillance and anti-theft also generate a lot of data, which means that a full set of big data categories (volume, rate of arrival and variety) is formed at the border.
- In the case of seismic systems used for data acquisition, projects include "converged" systems placed in ISO containers to acquire and reformat seismic data, potentially up to 10 petabytes of data. Due to the remote locations in which these intelligence systems operate, there is a severe lack of bandwidth to move data from the last mile border to the data center across networks. Therefore, service companies literally send data from the border to the data center on tape, optical or durable magnetic storage devices.
- Operators of brownfield plants, where thousands of events and dozens of “red alarms” occur every day, want to operate more optimally and consistently. However, low data rate networks and the near absence of repositories to collect data for analysis in factories suggest that something more fundamental is needed before starting a basic analysis of current operations.
It certainly makes me think that while public cloud providers are trying to bring all this data to their platforms, there is a harsh reality that needs to be dealt with. Perhaps the best way to classify this problem is to try to push an elephant through a straw! However, many of the virtues of the cloud are essential. So what can we do?
Transition to Edge Cloud Systems
Of course, there are already (industry) optimized solutions on the Hitachi market that enrich data at the edge, analyze it and compress it to the smallest usable amount of data, as well as provide business advisory systems that can improve processes associated with edge computing. However, my conclusion from last week is that the solutions to these complex problems are not so much about the widget you put on the table, but about the approach to solving the problem. This is truly the spirit of Hitachi Insight Group's Lumada platform, as it includes methods for engaging users, ecosystems, and, where necessary, provides discussion tools. I was very happy to get back to solving problems (rather than selling products) because Matt Hall said, “I was happy to see that Hitachi people were starting to get the scale of the problem right” as we closed our summit.
So can O&G (oil and gas industry) be a living example of the need for edge computing? It appears that given the issues uncovered during our summit, as well as other industry interactions, the likely answer is yes. Perhaps the reason for this is so clear because edge computing, purpose built for the industry, and a blending of cloud design patterns are evident as stacks modernize. I think that in this case the question "how" deserves attention. Using Matt's quote from the last paragraph, we understand how to push the cloud computing principle to edge computing. Basically, for this industry, we have to make “old-fashioned” and sometimes personal contacts with people who are involved in various parts of the oil and gas industry ecosystem, such as geologists, drilling engineers, geophysicists, and so on. With these interactions to be addressed, their scope and depth become more apparent and even compelling. Then, once we have the execution plans in place and put them into action, we decide to build edge cloud systems. However, if we sit in the middle and just read and present these issues, we won't have enough understanding and empathy to really do our best. So, once again, yes, oil and gas will give rise to cloud edge systems, but it is understanding the real needs of users on the ground that will help us determine which problems are of paramount importance.
Source: habr.com