"The only thing Microsoft should be worried about right now is how bad their publicity engine is. I mean, they're just letting Google walk all over them. They need to bring out their big guns and call BS on Chrome as an OS. Or, at least, force any rapid, fanatical anti-MS person to exclusively use browser-based apps for a week. "
-- David Geller

Our top 10 favorite headlines about Google Chrome OS

Mass High Tech

Microsoft's New OS

VIEW ALL

Bill Gates Watch

VIEW ALL

FROM THE TRENCHES

Guest Post: A case against the cloud

By Matt Seidel on November 7, 2008 at 6:26 AM PST

Comments (11) | Permalink

Cloud computing | From the trenches | Amazon.com | Google | Microsoft

Matt Seidel

Cloud computing is the latest buzzword for the big technology companies, the most vocal being Google, Microsoft, and Amazon. They’re fighting tooth and nail to be the pioneers and prove to the public they can be trusted with their services and your data. But I can’t honestly see why a developer or small/mid-size business would sign up.

You can point to all the glitches that major “cloud” services had in 2008 alone. (CNET’s Rafe Needleman had a great roundup of the 10 worst web glitches of 2008 so far.) Amazon’s S3 outage probably was the biggest. S3 is just one of the services bundled in what the company dubs Amazon Web Services. It sells storage space accessible via the Internet to many Web 2.0 companies that developed sites and services based on this infrastructure. Back in February, this service went down -- bringing offline all these companies that require the data stored on Amazon’s servers.

And we all know what happens when sites go down: they begin to lose money, customers or both.

The list continues with Google Docs, MobileMe and Gmail – all experiencing extended outages.

This is important to think about. As we learned in the past with these companies, when outages occur, they quickly seal their lips, leaving the customer in the dark on status, the cause, and whether their data is safe. In the internal IT world, if the corporate email server or production web site goes down, you can bet an email from some director or VP is asking for specifics – a full report on what happened, why it happened, what’s the damage, and what’s going to happen to prevent it the next time.

Good luck getting that kind of detailed information from Microsoft, Google, or Amazon.

Amazon’s core business is to sell goods. Google‘s core business is search and advertising. Microsoft’s is traditional software. Are these really the companies we should be relying on to offer a highly reliable cloud infrastructure? Not to mention questions of security and data control, and the risks of storing valuable intellectual property outside your company’s firewalls.

I’m not trying to dissuade anyone from using cloud services in the long run. I believe in the possibilities the cloud can provide.

But I think these companies are trying to lock decision-makers into proprietary systems that may continue to cost them more over time when compared to hosting in house.

We’re still in the early stages of providing computing over the Internet and maybe, over time, some of these concerns will become obsolete. But for now, I’ll be staying away from the cloud.

Matt Seidel, a Windows system administrator, blogs at www.atebymonsters.com. Guest posts are the opinions of their authors, and don't necessarily represent the viewpoint of TechFlash or its staff.

Bill said:

on November 7, 2008 at 8:01 AM

This is a really weak argument, basically just slinging FUD instead of actually comparing the in-house vs. cloud approaches.

Sounds like a sysadmin trying to save his job.

Report Abuse

Matt S. said:

on November 7, 2008 at 8:28 AM

The cloud approach locks you into a proprietary system that you have no flexibility with.

You provide services that relay on this cloud infrastructure. What happens if you clients needs change and require you to drastically change your development . . . maybe having a requirement to use Oracle or MySQL for the backend, or to host PHP based webfiles. Being locked into these proprietary system makes these changes far more difficult.

But more importantly, doing thing in-house allows you to have the resources on staff to set SLA's, and be on the spot to fix issues at that very moment. No need to talk to some first level support call center and then go through a long escalation process. All the outages that took place in 2008 alone is basis to give it a second though.

But I think knowing how your data or IP is controlled outside your firewall is pretty important FUD to sling, in my opinion.

Report Abuse

Mike Mathieu said:

on November 7, 2008 at 9:10 AM

Right now, there are two unique advantages to cloud services: 1) ability to handle peak loads in a capital-efficient way, and 2) ability to offload many administration costs or turn them from fixed into variable costs. What you give up to get these benefits is control. But that's also true when I buy electricity from Seattle City Light instead of generating it myself. And yes, some people get backup generators. I think that's probably the analogy to Microsoft's software+services model vs. a cloud-only approach.

Walk Score uses Amazon S3 for content and EC2 for map tile serving. Our soon-to-launch API is built on Google AppEngine. And of course our core app uses the Google Local Search API and Google Maps via AJAX. Walk Score wouldn't exist without cloud services.

That said, the issue of lock-in is huge, huge, huge. Particularly when it impacts your application architecture - like AppEngine - you have to realize you're making a long term investment.

Report Abuse

Drake said:

on November 7, 2008 at 9:15 AM

The cost of cloud computing outages pale in comparison to the costs of managing hosted legacy software. Staffing, bugs-fixes, forklift upgrades, firewalling, support contracts, hardware, power, cooling......the list goes on and on.

Like all utilities, cloud computing services will evolve to deliver the QoS needed by all businesses at the lowest cost per CPU.

Report Abuse

kayvaan said:

on November 7, 2008 at 10:44 AM

Matt;

You make some good points, but typical of this type of argument, you don't really do an apples to apples comparison.

First, it's not like local, on-premises apps don't go down. They go down all the time.

So you should compare uptime of cloud services not to an ideal of 100% uptime but to the ACTUAL uptime you get locally.

Second, you have to factor in the incredible cost savings of not having to do patching, upgrades, systems management, backups, etc. All the server baby-sitting that is really not value-added to any company.

I see so many of these contrarian view-points highlighting service outages with no additional context.

I'm not saying that cloud services are fully baked right now. But when you compare on-premises services to cloud services for your company, you have to look at the whole thing, end-to-end.

Report Abuse

Matt S. said:

on November 7, 2008 at 10:57 AM

kayvaan,

Totally understand . . . Like I said in my article I'm not trying to dissuade people, but hopefully spark them to really sit down and iron out the positives and negatives. Jumping to the cloud like some small businesses are doing, because it's the latest buzzword, will result into problems.

"First, it's not like local, on-premises apps don't go down. They go down all the time."

I hope my piece didn't give too much of an indication that in-house apps/servers don't go down. The point I was trying to make is when things do go down in-house you have the power/control to know the who, what, where, and why . . . not to mention the staff able to fix it on the spot and not rely on a third party, or call screen - I mean call center.

I know these third parties are not going to be as open when stuff really hits the fan.

But like I said, people just better read between the lines before making the jump. Like myself and Mike Mathieu said, you are locked in.

Report Abuse

E said:

on November 10, 2008 at 11:38 AM

Some companies run far more reliable services than Google or Amazon can provide, with 100% uptime. My employer is one such place. We've never had a microsecond of downtime in two years of running, and we can't (securities processing). If we did, there's a good chance we'd lose a lot of business.

Were there a problem, we'd want our systems to degrade in specific, known ways that we've thought about ahead of time. For example, in some failure scenarios, we'd want to continue running in a degraded manner, but with no customer visible impact. In other, more massive scenarios, we'd need to failover to a disaster recovery site, which would require customer involvement.

For Google and Amazon, you're "just another customer". Per their service level agreements, I'm sure there's all kinds of language which gets them "a free out" when it comes to providing services, not losing your data, etc.

If uptime is one of your company's core competencies, it's doubtful that Google or Amazon can do it in a way that will truly satisfy your goals. If you're "just a Web 2.0 company" (no disrespect, different goals, different challenges), you may be able to handle this, but for a lot of us, we'd never even consider using it.

Report Abuse

Rob Nachbar said:

on January 24, 2009 at 9:45 AM

As a small biz owner, the Cloud has become an increasingly critical part of my business. And I suspect this is true for many other small and medium sized businesses. I rely on Google Apps to collaborate with clients, Mozy to automate my back-ups, LiquidPlanner to manage project schedules, and the list goes on. In most instances, these services are provided for free or at a nominal (and predictable) monthly subscription fee that enable me to effectively compete with larger companies that have vast resources at their disposal to build and manage such systems. While there has been the occasional hiccup in service, it usually means just waiting for a minute or so and trying again. For large enterprise organizations I can see why they might be hesitant to migrate critical business services to the Cloud for reasons of compliance, security, and control. But for small and medium sized businesses, it's a no brainer.

Report Abuse

Anony206 said:

on April 11, 2009 at 12:24 PM

So the "Cloud" needs to be evaporated. At sea-level a "cloud" is called "fog" - that is what you get from most cloud providers. You get from point a to b, but you can't really see much. Insight into the "cloud" is needed, so lets call it what it is, "Managed Services on a Platform". This platform is managed as a whole, with redundancy in as many places as possible. Google doesn't buy 1U servers - it builds the cheapest components into computing nodes of hundreds of motherboards, CPU's and disks. If one fails, it either takes itself out of the node, or is killed by automated controls.

We complain about "legacy" this and that - well last time I checked most business software is "legacy" in its design. There is no way to stitch together cloud implementations across vendors. SAP and other big accounting / ERM apps will never be moved to the cloud, since the data the use is highly confidential!

You should hope the cloud you're flying into never disappears since you'll be left fully exposed.

The cloud is great for flexible customer facing tasks, and off-loading non-critical bandwidth and functions. Anything critical needs to be in an environment where a failure is detected, diagnosed and resolved with full communications with the business leadership.

I'm a big VMWare VI user, and building clouds is what I do. The clouds I build for customers can be relocated as needed, with no lock-in issues. They can deploy ANY application, or convert their existing infrastructure to be part of the cloud. I have clear point in time backups which notify me that they completed, and the monitoring of system resources to free up cycles for heavy users.

Report Abuse

Anonymous said:

on April 18, 2009 at 3:58 AM

Microsoft products did not fail.
So, use only Microsoft products.
Forget Google & Amazon...

Report Abuse

everRun said:

on April 26, 2009 at 11:21 AM

New everRun 2G bring Continuous Application Availability to the Mainstream

Marathon Technologies Launches Industrys First System-Level Fault-Tolerant Software for Virtual Environments and Comprehensive Windows Availability Platform

http://www.marathontechnologies.com/launches_industrys_system_level_fault_tolerant_software.html?newsid_press_release=985

Report Abuse

Todd Bishop's Biography

Todd Bishop is co-founder and managing editor of TechFlash. He has covered Microsoft and the technology industry for more than five years, most recently as a daily newspaper reporter and blogger based in Seattle. READ FULL BIOGRAPHY

206-876-5466

Recent Sponsor Posts

Gnomedex: Tickets Available Now for Puget Sound's Premier Tech Conference Aug. 20-22

Join tech enthusiasts from across the country for the leading tech conference in the Puget Sound area, Gnomedex. It's the technology conference of inspiration and influence, hosted by Seattle-based blogger and digital enthusiast Chris Pirillo. Gnomedex is geared toward anyone with an interest in technology, and openness for sharing ideas and learning from others.

MORE INFORMATION

SynapticMash: A Microsoft BizSpark startup using technology to increase student achievement

How do successful startups like SynapticMash survive in today’s economy? Innovation gives them a leg-up; as does help from programs like Microsoft BizSpark. BizSpark is a program for software startups that provides development tools and technologies to build applications and server licenses to host them.