Say what you will about Windows, but anyone who remembers the old days of DOS where each app had to maintain its own drivers for printers, modems, etc. or even back to CP/M where every terminal, keyboard and disk drive needed individual support, we see that Windows was a very attractive way to abstract applications from the hardware. In addition, it did provide a common user interface (which was attempted by many in the non-graphical DOS era) as well as multi-tasking, which also had been attempted in primitive ways before (remember TSRs and Quarterdeck DesQ ?)
The best thing about Windows was that in being nearly ubiquitous, it drove down training costs, and allowed (for the most part) a common OS for both clients and servers. Windows NT squashed its opponents such as NetWare and Banyan, despite the fact those offered many superior core technologies.
Linux still serves as a very viable alternative, however its fragmentation has cost it market share compared to Windows.
Taken further, the incredible success of VMware (and the emergence of alternatives like Xen, Solaris containers, etc.) were inevitable as hardware advanced to far exceed its exploitation by a single OS per machine model. Virtualization created its own set of needs (live migration, virtual networking, Physical to Virtual conversion, storage virtualization, etc.) unique to the new capabilities. As Windows abstracted APPLICATIONS from hardware, Hypervisors take the next logical step and abstract the OS from hardware.
Granted, a hybrid approach to virtualization such as Solaris containers DOES, in fact, combine the OS and virtualization stack, I suspect the fact that is is a proprietary technology would naturally limit its market share to a large extent. It is the ability to support multiple OS flavors and versions side-by-side that is a great part of virtualization's appeal, and 3rd party hypervisors not tied to any OS will likely continue to do this best.
It's kind of funny, that the management features unique to virtualization arose to solve problems that never before existed, but in the end, the benefits of virtualization end up an easy-tradeoff for the additional necessity to learn some new tools, and architect in a way to best take advantage of the technology.
I don't see the hypervisor taking over the role of OS anytime soon. It's just my opinion, but keeping it simple will pay dividends. I can, howewer, see the role (or importance) of the OS diminished significantly as "virtual appliances" evolve.
Check out www.rpath.com (they seem to have been doing this the longest) !! If you could acquire an app running on an open-source Linux distro that fulfilled all the needs of your organazation, you would find little excuse to select a similar app that requires a full Windows license.
In the end, I think SUPPORT will be the big issue for most. I recall, when I worked at HP, our big selling point was that we supported GUEST Operating Systems as well as the underlying hypervisors. That was a HUGE selling feature, as Operating Systems vendors (at the time) chose not to support their wares on virtualized platforms.
Cloud Computing has the opportunity to take that question off the plates of end-users.
On Wed, Dec 24, 2008 at 12:13 PM, <barbara@p..com> wrote:
All,
This thread started under topic of: Role of Windows Appliances and
Cloud
I opened a dedicated topic.
THE QUESTION (Response from trimark [m_cathcart@...co.uk] follows.
What is Future of conventional full service OS in the world of
virtualization.
OS's run on top of Hypervisors. And...Hypervisors are enablers of IAAS
because, as we all know, most existing software systems dwell
comfortably in OS's level of granularity, and are not at home in SAAS
or PAAS.
I am looking at the question from a broad perspective.
The complexity that is introduced by black box hypervisors troubles
me. I feel the solve a problem already solved by IBM and other big box
vendors decades ago (VM/CMS).
So, with that in mind, the whole thing begs the question, what is the
future of the full service OS?
Do they morph into a meta OS, that can function in a manner like a
hypervisor OR full service OS or both (in essence something like a
mainframe OS).
Barbara Bour
RESPONSE M_CARTCART
Barbara
You ask what is the future of full service OS's and what next.
I'd say this, that its pretty obvious that services will be
increasingly provided by the hypervisor and virtualztion layer, esp
[ecially networking and storage, but also possible virtual memory, I/O
etc.
If the OS running on the hypervisor provides the same service and
doesn't do this WITH the hypervisor then it will be inefficient, and
generate overhead. There are a few possible solutions to this, one is
for the OS NOT to provide the same function, thats unlikely with full
OS's, or as part of initialization, they recognise they are running in
a virtualized environment, and cede that function to the hypervisor.
The latter makes more sense, but requires compatibility between the
hypervisor function and the OS funtion at the application level. For
higher level functions and new technology, thats easy. For lower level
functions this is less likely. It also removes much of the opportunity
for the OS to provide differentiated services.
Back to your question, what of the future for full service os?
I'd predict that most will do their damdest to link themselves with
their own hypervisors. In that way they can continue to provide
differentiated services that allow them to continue to be sold at a
premium. However, these OS's will generally lag behind emerging
composite OS's, where the hypervisor and the OS are made from a
networked, interconnected set of services, with little generation or
overhead between their services.
It would be interesting to know where/how you think IBM solved this
decade ago? As I see it, the only place IBM really solved this was in
VM/CMS. Where there was a strict hypervisor/virtualization layer that
had unique calls for functions, and the CMS OS which developed into a
purely virtualized OS and couldn't run without VM.
Other IBM implementations including AIX on Power did this to a lesser
degree, but really still are full function OS's that use
virtualization sparingly as their host.
We are likely to see an effort obsfucate the OS to providing grouped
higher level services that are interfaces into homogenous full service
OS's, these will provide a single point of automation, management,
etc. as well as scheduling and recovery. In this way, the full
function OS remains and the function that you are looking for from a
cloud is provided by a layer on top of the OS, rather than underneath
the OS at teh hypervisor layer. From a management, operations
perspective there is little differtence. From an apllication and
operational efficiency perspective there is a significant difference.
--
Cheers,
Jan
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google
Groups "Cloud Computing" group.
To post to this group, send email to cloud-computing@googlegroups.com
To unsubscribe from this group, send email to
cloud-computing-unsubscribe@googlegroups.com
To post job listing, send email to jobs@cloudjobs.net (position title, employer and location in subject, description in message body) or visit http://www.cloudjobs.net
To submit your resume for cloud computing job bank, send it to resume@cloudjobs.net.
For more options, visit this group at
http://groups.google.ca/group/cloud-computing?hl=en?hl=en
Posting guidelines:
http://groups.google.ca/group/cloud-computing/web/frequently-asked-qu...
This group posts are licensed under a Creative Commons Attribution-Share Alike 3.0 United States License http://creativecommons.org/licenses/by-sa/3.0/us/
Group Members Meet up Calendar - http://groups.google.ca/group/cloud-computing/web/meet-up-calendar
-~----------~----~----~----~------~----~------~--~---
Posts
No comments:
Post a Comment