Mailbag: Much Ado About the Cloud, .NET vs. Java, More
Doug wrote about Larry Ellison's criticisms of cloud computing. Here are some
of your own thoughts:
Cloud, shmoud. Yes, the cloud is way overhyped. I'm already tired of
hearing about it. Isn't this just another name for client/server computing?
Ho-hum. Been there, way past that.
Oh, and the day I would trust our company's data and/or applications to the
cloud is the day the entire Internet decides to take a dump and I would be
shown the door. No thanks.
I think that anyone dismissing cloud computing as hype doesn't know what's
going on around them. We're going through another swing toward "mainframe"
computing, but this time instead of using thin clients to access user sessions
on beefier servers, individual servers are being virtualized on large servers,
and the new thin client is the browser.
I would recommend that people look into hosted cloud offerings such as
Mosso.com, Amazon's EC2 and others to get an idea of what utility cloud computing
really is. You pay for what you use, and your environment scales dynamically
to meet your usage needs. You no longer need to spec out individual pieces
of hardware for hosting certain applications. You just put your applications
online and go. You pay for the disk space, bandwidth and CPU time that you
use. Cloud computing is service on-demand. Many SaaS providers are hosting
on these types of platforms to dynamically scale their application as they
add subscribers. The mainframe is getting much smarter. We apologize if it
has a catchy name.
A recent survey suggests that .NET
is making gains against Java. Readers share where they stand:
A lot depends on what you are doing and what you need to do. We use both
.NET and Java; some of the tools we use are .NET-centric and some are Jave-centric.
The tools perform very well and play nicely with each other. Here's my take,
though: If Microsoft doesn't start lowering the prices, I think it's going
to end up pricing itself out of the market. Yes, its tools are well-polished,
but it's not like the average programmer can pick up a copy and play with
it. You used to be able to do that with VB and still afford lunch, but that
day has disappeared. Plus, the toolset has gotten a lot more complicated and
more pricey. I can download any number of IDEs for Java development and pick
the one that suits me. With .NET, I'm pretty much limited to MS -- especially
if what I'm doing is mission-critical.
I used to have an MSDN subscription, but I honestly can't justify the
cost to my boss. In fact, I'd rather have them spend the money and give me
a standalone SAP system, which is another thing I develop for. I don't even
need to worry about .NET vs. Java in that environment; I just use what they
provide and it all plays nicely together. So if you're developing the latest
tool for Symantec, then yeah, .NET and MSDN is the way to go. But if it doesn't
matter what builds your app, I think MS has a LONG way to go to justify the
As far as .NET vs. Java, we are a custom software dev shop in Austin,
Texas that USED to do both Java and .NET -- but we have not had a single customer
requesting Java in more than three years! Every project where the customer's
RFP specifies the base technology, it has almost always been .NET (and occasionally
PHP on Web projects). And where they have not specified, I cannot remember
a single project that ended up being a Java project. The only Java developers
I still know are some guys who are evolving a five-year-old application for
Pitney Bowes, and even they are starting to move to .NET for future versions.
As for us, we are now pretty much exclusively a C#/.NET house.
A few more of you chimed in on VMware and its plan to create a datacenter OS:
VMware's plans for the datacenter OS are great ideas if it can make them
work. This would allow us to use some of our older hardware for the processing
power that otherwise have gone to the scrap heap. It could also lead to faster
results and provide a better user experience. In terms of moving this eventually
to the cloud, I believe there are going to be many reservations and there
will have to be some high-level SLAs to justify this move.
Not so long ago, I worked for a litigation support company with a close
relationship with EMC, a relationship that was doomed because EMC eventually
decided to be a competitor and the litigation support company was acquired
by a search engine vendor without a single clue. During the EMC days, however,
we had VMware forced down our throats at every opportunity, first on our development
machines and later with the enterprise edition forced into our datacenter.
I hated it on the development machines, both because of its effect on peformance
and because we repeatedly lost significant amounts of development due to virtual
machine issues. (I eventually refused to continue using VMware on development
machines, switching fully to Virtual PC for improved performance and greater
stability.) Use of enterprise editions of VMware in the datacenter proved
nearly catastophic. The servers on which it was deployed suffered severe performance
degradation and increasingly erratic behavior. It was yanked from all production
servers shortly thereafter.
In my opinion, VMWare enjoyed success in the virtualization market as an early,
major player with few competitors, a level of success that's quickly fading,
in part because it lacks the technical and business expertise to effectively
compete with later-arriving players and because virtualization is a technology
best embedded in the core operating system. I fully expect Microsoft to reduce
VMware to "also-ran" status in the virtualization market -- and
I look forward to it doing so.
Philip shares his thoughts on exactly what Microsoft's High
Performance Computing (HPC) Sever is capable of doing:
The whole HPC issue is a little muddied since the cores of current desktop
and server machiens are plenty busy due to them being dispatched by the large
number of concurrent threads running on the typical machine. Sorry to disagree
with you -- and also Microsoft -- on this issue. Microsoft's HPC push is designed
to deal with the next generation of processors that will have MANY more cores
(dozens, hundreds), and there may not be enough work via a typical multithreaded
application and operating system to keep the cores busy
In essence, the HPC push is an educational and technical push to get developers
to break up their applications into smaller subunits of execution (which is
not possible or practical with current compiler and OS technology). HPC is
also a necessary evolution of Microsoft's .NET technology to provide it with
better scalability and also some unique distinctions so that it is a more
attractive platform for high-end/high-performance applications.
And Fred thinks there's more to Microsoft's recent $40
billion stock buyback than meets the eye
I wonder whether this is Microsoft buying back assorted shares of lots
of small investors on the open market...or Bill and Steve cashing in by having
Microsoft buy back some (or all?) of THEIR shares before the market implodes
We want to know what you think! Leave a comment below or send an e-mail to
Posted by Doug Barney on 10/02/2008 at 1:16 PM