The Promise and Reality of CAD on the Cloud23 Jul, 2014 By: Robert Green
How has the cloud materialized for CAD users and managers?
Storing Data on the Cloud
Years ago, when users questioned the security of placing their data on a cloud server, they were usually met with this argument from software companies: "Don't you use online banking? Don't you buy things with a credit card over the Internet? Why worry?" The implication was that someday soon, all our data would be on "the cloud," and the traditional data center/server would be a thing of the past.
Today, however, we know better. The news is full of stories that prove we should worry: the Target credit card hacking case, Dropbox's hugely embarrassing lapse in password security, millions stolen from bank ATM networks, outright theft of intellectual property by Chinese hackers. It is glaringly obvious that the cloud isn't always a safe place to store sensitive data. Given all these problems, what CAD manager will recommend to his or her senior management teams that all the company's CAD software and models/drawings should be stored somewhere other than the company's own data center?
Conclusion: Data security is of paramount concern to businesses, and all the evidence points toward an increasingly insecure digital infrastructure. In addition, since getting to the cloud means running over the Internet, in most cases I believe companies will to continue to keep their data on their own servers.
Speed and Bandwidth
Years ago we were told that high-speed, fiber-based Internet service would be ubiquitous — the whole world would be wired for it. Statements such as, "With the cloud, it won't matter where you are working," were widely repeated. As of today's writing, I've found neither of these prognostications to be even remotely true. Don't get me started on trying to do business at a construction site in the Philippines or an Air Force base in Italy. I can't even get a solid LTE/4G connection in many rural areas of the United States!
What hasn't changed is that most users experience "the cloud" by connecting to it via some sort of broadband Internet or LAN. What that ultimately means is that your experience with the cloud is almost totally dependent on the speed and bandwidth of your connection to it. And, while software has changed a lot in the past five years, the speed at which computers all over the world can talk to each other hasn't increased much. It doesn't take a rocket scientist to figure out that more users accessing more data on more cloud applications with nearly the same network bandwidth as before will contribute to a big slowdown in performance.
Conclusion: Moving to the cloud only makes sense if there is the bandwidth and speed present to do so. Right now neither is a given, so migration to the cloud continues to be a slow process.
Renting CPU Cores
No matter how nice your workstations are, there is always a newer, faster model, right? And when you have to complete a lot of rendering or analysis jobs at once, you never seem to have enough computing power to do so quickly.
Enter a concept that wasn't really discussed during the introduction of the cloud: renting computer cores. Wouldn't it be nice if you could get access to a couple of 12-core machines equipped with 192 GB of RAM to run your rendering jobs for a week or so? Cloud-based CPU farms make this scenario more and more realistic, as service providers configure high-powered workstations with popular software that you can use for short periods for a reasonable fee.
There are multiple advantages to this scenario:
- No machine ownership. That's the vendor's problem.
- No IT headaches. No configuring graphics cards, network interfaces, etc.
- Conserving cash. Pay just for what you need, at a fraction of the price of full-time ownership.
- Sharing is an option. Your Arizona office could run analysis one day, Sydney the next day, and New York all next week. When everyone's done, just terminate the rental.
- You own the data. Just upload your software/models, run the analysis/rendering, and download your results.
- Security risk is minimized. Your data is only on the cloud while the analysis is running, rather than permanently residing there, so any exposure you do have is greatly reduced.
- Speed and bandwidth issues are minimized. Because initial upload and downloads are the only time when these issues matter — rather than continually moving application data back and forth between the cloud and the user — CPU rental is no more data-intensive than the current uploading/downloading we experience today.
While the cost of CAD workstations is minimal for most users, the really high-end machines used for only occasional analysis tasking are still expensive. When renting the power of these high-end machines is as easy as uploading/downloading some files, why would we continue to suffer through the drudgery of processing on old hardware anymore?
Conclusion: Renting CPU cores is an area of cloud computing that offers real value for businesses, and poses a minimal security risk. While the cloud was conceptualized as a storage environment, it may turn out to be CPU rental and sharing that serves CAD needs the best.
Summing Up — for Now
In the next edition of the CAD Manager's Newsletter I'll share some technology advances and strategies that I see companies using to create their own private/hybrid cloud environments. In the meantime, I hope you'll use these concepts to help formulate your own cloud strategy and start the discussion with your senior management and IT personnel to make sure everyone is on board.
I'm keenly interested in your thoughts about CAD on the cloud. Please send your thoughts to firstname.lastname@example.org or swing by my Facebook group "CAD Managers Unite!" and leave your feedback there. You may even see your name in the next newsletter! Until next time.