Bruce Friedman, of Lab Soft News, ruminates on the coming of the serverless company and, in an interesting twist, suggests that the emerging utility computing industry may end up organized along vertical lines, with the big utilities having subsidiaries specializing in providing services to different industries. Freidman’s particular interest is the health care business:
Healthcare IT has traditionally lagged about a decade behind computing in other corporate sectors. However, it seems to me inevitable that hospital EMRs, LISs, RISs, and other information systems will eventually run on rented servers rather than on hardware owned and managed by the hospitals themselves. In time, the potential cost savings will be irresistible. It also occurs to me, however, that hospital executives would never be able to tolerate using, say, an Amazon server farm or “computing cloud” because of the perception that healthcare computing is different than corporate computing or e-commerce.
Therefore, I envision specialized companies, or even subdivisions of existing companies like Amazon or Google, that will have “healthcare” in their names and will serve only the healthcare industry. In this way, they will be able offer an array of services specifically tailored to the needs of their healthcare clients. Paramount, of course, will be non-stop computing, rapid response time/disaster recovery, and iron-clad data security/confidentiality. Frankly, these demands will not be that different than those demanded by many current e-business customers but I am sure that the perception of specialized services will be comforting to the hospital executives.
The verticalization of the cloud would provide marketing benefits, as Friedman notes, while also providing a possible means of addressing issues of information security crucial to industries such as health care and financial services. It might provide a regulatory stepping stone between private systems and a shared grid. Even if the underlying processing grid was shared, vertical offerings would also make sense for industries characterized by highly specialized applications, like retailing.
UPDATE: A commenter on Friedman’s post notes that the vertical health-care cloud is already well established, at least for the critical function of data storage and retrieval: “InsiteOne, a healthcare imaging storage provider, currently has over one billion image files stored for over 400 hospitals who pay a fee for each stored file which can be accessed at any time for referral. The vertical cloud continues to get bigger with time and healthcare is joining this cloud in even greater numbers. You can look at server utilization now in the same light as your electric, water, and other utilities services.”
I see a lot of people making conjectures about ‘cloud computing’, or at least, this year’s version of ‘cloud computing’, and I wonder how many have ever worked for the IT department of a larger corporation, or actually directly worked in the technology in the first place.
What, on earth, makes you all assume that purchasing and maintaining servers is the big cost factor in IT systems? How do you think most of the major corporate applications work? Throw up a web page and enter patient information?
But the utility model doesn’t just address server costs; it also addresses (or can address) software licensing and maintenance costs and (the biggest current IT expense) labor costs as well as electricity, real estate, and other secondary costs related to private data centers and systems, not to mention the attendant risks. Also, because it makes the sharing of data much easier, it is extremely well-suited to health care, where private, incompatible systems perpetuate rather than solve the problem of fragmented patient records. You’re right to caution against oversimplification, but you’re oversimplifying a bit, too.
Software licensing concerns — that’s not the server problem. The clients are where the licensing costs come into the picture in a corporate environment. Licensing costs for something like Oracle — do you really think a lot of companies are going to want to do their SQL processing over an external pipe?
The bigger companies with bigger data centers are probably right next door to the centers serving the clouds — whether you pay for the electricity directly or indirectly, you’re still paying.
Sharing data easier? Location and pipe has never been the issue with sharing data. Never, ever, ever, and I’ve been working with this since I helped Boeing with the old POSC effort almost three decades ago. It is initial structure of data, and format of data transmission that’s always been the issue. Location of machines — big whoop.
Locating health service data on a cloud won’t make the data more sharable, if structure and format are incompatible, or the organization is unsure about what they can and cannot legally share — much less being concerned about the shared parties own internal security, and liability if they’re data monkeys.
Data sharing issues have been around a whole lot longer than the web, and especially “Web 2.0” and it’s fluffy clouds. Of the many problems, one that didn’t exist was the physical location of the machines.
I’m not simplifying, I’m speaking from experience.
I agree with Shelley that there is still a question hanging over utility computing for large enterprises. That said, we are seeing strong growth in Managed Services, SaaS, etc., which also fall under the broad definition of “cloud computing”. So in cases where the enterprise gets a whole “package” of value it makes sense for them.
This could be especially true for verticals such as healthcare where you can add specific services, such as HL7 processing middleware (or whatever the standard of the day is in each industry).
From our experience at GigaSpaces, there are a few scenarios where utility or cloud computing also make sense for large enterprises, and those are:
1) Testing phase — if you want to test a large, distributed environment and aren’t quite ready to invest in the hardware resources
2)Handle inconsistent peak loads — some of our customers are using our product on EC2 to handle unexpected loads. They don’t want to provision hardware that would be idle most of the time and can use such services on-demand for better economics.
And then there are start-ups. For many of them it makes total sense to start running on a cloud computing service, and only invest in hardware and software as the business grows (that’s why we are offering our Start-Up Program.
Geva Perry
“A commenter on Friedman’s post notes that the vertical health-care cloud is already well established, at least for the critical function of data storage and retrieval”
Image storage does not require any agreement as to format; not does it require any agreement as to how an exchange of data is to be made.
Offline archival storage to support large media files, such as the images this company manages, and to enable disaster recovery has been around long before the web, and is not the same as ‘a cloud’ for serverless computing.
Geva mentions some of the cases where such serverless environments exist, such as new startups, or for mirroring a system for testing (though thats more in line with just leasing a group of machines for a finite period of time).
Even InSite’s own About Us reflects the archival nature of the site:
“InSite One is the leading service provider of medical data archiving, storage, and disaster-recovery solutions to the healthcare industry.”
That’s not a cloud.