Healthcare Experts Question Open-Source Apps (LinuxInsider) [LWN.net]

Healthcare Experts Question Open-Source Apps (LinuxInsider)

Posted Jul 21, 2005 21:58 UTC (Thu) by AnswerGuy (subscriber, #1256)
In reply to: Healthcare Experts Question Open-Source Apps (LinuxInsider) by sfeam
Parent article: Healthcare Experts Question Open-Source Apps (LinuxInsider)

"It is at least possible that "no support for continuous upgrades" was a poorly-phrased way of saying "no support for patching or upgrading a running kernel".

This seems sufficiently implausible the we can safely consider it to be impossible.

Neither Windows nor any major versions of UNIX have this capability whereas there have been some unofficial patches (kexec()) for Linux that have explored the possibility.

In fact I think we can safely say that the whole statement was a red herring, an excuse to justify a foregone conclusion whose only real basis was an existing prejudice.

At this point the IT for large healtcare institutions is ensconced. The major concerns are political rather than financial or practical. There is quite alot of concern about liability, HIPAA and regulatory issues which are a fertile breeding ground for a culture of over-priced, under powered, egregiously complex software and service offerings which are long on vendor lock-in and very short on flexibility, robustness, or other features.

JimD

(Log in to post comments)

Healthcare Experts Question Open-Source Apps (LinuxInsider)

Posted Jul 21, 2005 23:31 UTC (Thu) by sfeam (subscriber, #2841) [Link]

There are more options in this world than Windows and Unix/linux. Some of the major systems here at the medical institution where I work are counterexamples to your conlusion of impossibility. They are OpenVMS clusters, and yes they can be upgraded on-the-fly because the cluster runs continually while individual machines are upgraded. You may scoff if you like, but this is indeed seen as an important feature.

Linux has a way to go before it can support this level of clustering, and I'm not sure it's really aiming in that direction anyhow. CPU hotswap is being worked on, so hardware upgrades may be possible on-the-fly, but so far as I know this is a separate issue from continuous operation across a kernel upgrade.

Healthcare Experts Question Open-Source Apps (LinuxInsider)

Posted Jul 22, 2005 0:30 UTC (Fri) by huffd (guest, #10382) [Link]

OH PLEASE!

Not another tired old "Linux can't myth"!!

Geeze these people have been doing it for AT LEAST five years:

http://www.scyld.com/platform_overview.html

And stop this "Linux can't" crap! If you can imagine IT, someone else has done IT, just because you've never heard of IT, doesn't mean that someone else hasn't already made IT, the cornerstone of their business!

Got IT?

Geeze..

Scyld is a master/slave system.

Posted Jul 22, 2005 1:04 UTC (Fri) by xoddam (subscriber, #2322) [Link]

It's not clear to me that the Scyld system allows a live kernel upgrade or migration of the 'master node' -- I'm wading through white papers and I haven't found it. This is a supercomputing cluster rather than a symmetric load-balancing database cluster, which is what medical information users would require.

Master Node Redundancy is a Future Enhancement

Posted Jul 22, 2005 1:18 UTC (Fri) by xoddam (subscriber, #2322) [Link]

Got it, in the Future Enhancements section of the technical overview pdf:

Master Node Redundancy — Scyld is planning to update the operating system to support multiple master nodes in the cluster. This feature will allow the transparent switchover to a backup master node in the event that the original master node suffers a failure.

So it's a planned feature, not an exising one.

Continuous operation across a kernel upgrade

Posted Jul 22, 2005 0:47 UTC (Fri) by xoddam (subscriber, #2322) [Link]

Process migration tools exist for Mosix clusters which allow
a "single system image" to continue to exist while individual
machines are taken down and brought back up. I'm not sure if
Mosix allows the machines to be running incompatible kernels
though.

Xen or VMWare would allow the clustering tools to be used
among partitions on a single machine, so you don't need multiple
physical machines to do the trick, but without process migration
there's no way to upgrade a running kernel while applications
keep running.

Continuous operation across a kernel upgrade

Posted Jul 22, 2005 6:33 UTC (Fri) by rqosa (guest, #24136) [Link]

> I'm not sure if Mosix allows the machines to be running incompatible kernels though.

It doesn't.

Furthermore, I doubt that SSI clustering is necessary for these kind of applications. Probably the Linux-HA project would be more relevant. From the FAQ:

HA (High availability Cluster) - This is a cluster that allows a host (or hosts) to become Highly Available, that means if one node goes down (or a service on that node goes down) another node can pick up the service or node and take over from the failed machine.