Addison Snell recently wrote an article for The Next Platform called “The three great lies of cloud computing.” Snell points out that the marketing around cloud computing doesn’t always match reality. As someone who does marketing for cloud computing software, I just want to go on the record as saying Addison is absolutely….right. We’ve spent a lot of time on this blog, at conferences, etc. talking about the benefits of using public cloud services for big compute and big data. But we believe that a one-size-fits-all solution is never the right size. Public cloud services can be sized to fit many needs, but not every need.
If there’s one area where Addison’s article falls short, it’s that he only considers the raw dollar amount when talking about cost. Raw dollar amount is important, of course, but it’s not the whole story. As I said in response to a question at HTCondor Week 2016, it’s all about the value that cloud resources provide, not the cost. If you spend twice as much to run a workload in the cloud, but you get three times the value (e.g. due to faster time-to-results or the ability to run simulations at a finer resolution), that’s a net win.
Another part of the value discussion is the total value of your entire HPC environment: the mix of cloud plus internal resources. Many organizations have very specialized systems in-house (InfiniBand interconnects, high RAM, parallel file systems, etc) that are optimized to deliver maximum performance for key workloads. But the clusters also end up being used for other workloads that don’t need — or can’t use — the specialized hardware. Leveraging cloud to move the workloads that don’t need the specialized hardware into the cloud so that more jobs can be run on the optimized hardware makes the total environment deliver greater value.
“It’s not really a cost problem,” VP of Sales and Ecosystem Tim Carroll told InsideHPC, “it’s about access and utility.” We have seen a customer get rid of month-end analysis runs because they can get the capacity to do the in-depth simulations during their nightly operational runs. One customer cut a manufacturing simulation from months to hours by moving from a small internal cluster to an 8,000 core cloud cluster. In these and many other cases, using cloud resources resulted in a higher cost, but the value delivered was many times greater. When you’re considering the costs of using public cloud for big compute, remember to consider the benefits, too.