Cloud Computing Reversed
I always believe cloud computing is a computing, not the computing of the future. Its elastic and centralized nature allows greater level of sharing that was otherwise impossible within single organizations. It works great for anyone who has dramatic workloads and other cases. But it doesn’t work in all the cases.
Recently a new use case comes to my attention. It actually requires opposite way to cloud computing. You may have known recent developments in bioinformatics. With human genes are sequenced and analyzed, we can do a lot of interesting researches, for example, correlate a symptom or medical reaction to certain gene sequences. This requires a huge numbers of gene samples to be analyzed.
Genetic researchers/doctors around the world generate huge amount of data from gene sequencing machines. As you can imagine, the data is highly sensitive. Therefore, to be safe, it’s not saved in cloud without any pre-processing like encryption, de-identification, etc. Even so, some state laws require medical data must not saved outside of state.
Technically, gene data mining is a great use case for cloud computing. But without data in the cloud, your applications there don’t help much. Also, for human genes, majority sequences are the same. So it doesn’t make sense to upload all the data but the much smaller delta.
So the typical cloud computing model doesn’t work here.
The solution is a reversed cloud – meaning you have applications downloadable (maybe packaged as a virtual appliance) and running inside researchers’ organization for initial data processing. The processed data does not have identification and significantly smaller than raw data, and is good to be sent to a cloud. This approach addresses the privacy concern and saves bandwidth and time.
This reversed cloud model solves the initial data processing issue. It’s actually complementary with typical cloud computing model which can then do the large scale cross gene analysis.