Integrien blog | Integrien

VMware as Virtualization ‘MythBusters’

VMware continues the quest to host its own version of Virtualization MythBusters.

Following a review of the company’s own performance testing, which was recently presented to analysts, VMware now has updated information further busting common myths associated with going virtual.

In a post July 22, ZDNet’s Dan Kusnetzky outlines some of them including:

·    Applications in virtual networks do not have to run slower
·    Applications with high input/output, can, in fact, be potential virtualization candidates

We applaud the simulated performance testing by VMware, which Kusnetzky characterized as “reasonable” and “supportable.” We also applaud one of VMware’s conclusions that many of virtual machine performance issues boil down to improperly configured virtual networks.

This is largely due to companies rushing to go virtual without adequately assessing several things, including what should be virtualized and what should not be. Further, complexities such as how changing loads affect hypervisor performance must be considered.

Most system management tools, as we have proselytized before, require too much manual effort to resolve performance problems in virtual networks. Further, these tools are not proactive in nature. Managing performance in a virtualized system is without a doubt more daunting than managing performance in a physical one. Better tools, quite simply, are needed.

Our IT Forecast: Cloudy, Some Intermittent Showers

The recent troubles at Amazon are casting an unfortunate dark cloud over, well, cloud-based services.

For more than six hours July 20, Amazon’s Simple Storage Service went black (“elevated error rates” as Amazon called it), affecting some applications of 37signals, Huffington Post, Twitter, iPhone and others. As if Apple needed any more server problems.

A cloud computing service offers potentially unlimited bandwidth as well as sophisticated virtualization technologies. Just as the Internet has made downloading media more popular than buying CDs and software from stores, cloud computing is allowing many companies to source their computing power and storage from afar. 

As a result, cloud computing has become incredibly popular of late. All sorts of companies have gained traction in this space, including GoGrid, 10gen and EngineYard. And a key Microsoft executive was recently quoted by Dow Jones saying he wants to "aggressively facilitate" the re-engineering of Microsoft Office for cloud computing environments. He also said Microsoft expects a substantial portion of revenue in the near future to come from online applications.

However, as Amazon discovered, running complex applications in a central environment over the Internet presents all sorts of problems. One problem is security, since cloud vendors often outsource some parts of the hosted service. Then there’s stability. Since so many users depend on one service, it’s absolutely crucial to have the smartest technologies available to manage and monitor a cloud system. Cloud companies, and unfortunately their customers (and their customers’ customers), will continue to see more major outages – probably far more severe – without predictive IT solutions.

Cloud computing is here to stay because it’s powerful and makes sense for many applications. But we will ultimately see occasional dark skies for the foreseeable future.

Greene Out at VMware. Now What?

So, the IT world was taken aback July 8 when VMware CEO Diane Greene was ousted. Now, we in the virtualization industry face a bit of uncertainty following the unexpected departure of this pioneer.

Greene, who by all accounts led a fledgling startup focusing on an unknown technology in the late 90s to a company that now commands 90-plus percent of the virtualization market, was fired by the board amid lower-than-expected revenue forecasts.

It was later revealed by insiders that Greene was at odds with the board and with VMware parent company EMC for two primary reasons. The first: The board wanted someone with experience running a large corporation, not just an entrepreneur. The second: Increased competition from Microsoft. Forrester analyst Frank Gillett, quoted in InfoWorld, says VMware should have been shifting from pure virtualization to “higher vision, strategy, brand and value proposition” with news of Microsoft’s new Hyper-V technology.

We believe VMware will have to shift its business strategy to compete against Microsoft and other mega corporations entering the virtualization fray. We also believe VMware’s new CEO, Paul Maritz, a former Microsofter, understands that his new company will have to adapt.

One adaption will likely be around offering more dynamic – diverse – IT offerings, as Gillett alluded to. Another will likely center around creating more partnerships, which has historically been Microsoft’s forte.

Our view: The more partnerships, the better.

From iPhone to Firefox. Prevention is Far Better Than Reaction

Is it us, or are there more high-profile server crashes of late? And doesn’t it also appear that even as more consumers depend on the Internet for their daily lives, companies that have crashes aren’t being all that upfront or apologetic about them?

On July 11, Apple unveiled with great fanfare the iPhone 3G. Long lines were witnessed coast to coast filled with eager Apple technology enthusiasts. However, in what is being dubbed “the iPocalypse,” a major server crash delayed the launch by hours. Thousands of new iPhone customers, including InformationWeek’s Mitch Wagner, basically had a $200 brick in their eager hands that couldn’t do anything other than look pretty.

In June, downloads of Mozilla’s newest Internet browser, Firefox 3.0, was delayed following a – you guessed it – server crash. Following a massive publicity blitz to enter into the Guinness Book of World Records with the most software downloads in one day, the Mozilla site was rendered useless, giving cryptic messages such as “Service Unavailable.”

Over the past month, reports Data Center Knowledge, there have been outages for other services including Apple’s Mobile Me, Facebook, Google Docs, 37signals and LiveSide.

Some of these crashes are as costly as they are incredible. Almost everyone wants an iPhone. Couldn’t Apple have prepared better to have its IT resources optimized for mega traffic? And Mozilla was actually preparing to enter Guinness. It doesn’t send a very good signal that you can compete against Explorer and Safari when your IT operations can’t even handle your own promotion.

Worse, blogs everywhere are complaining about the poor way in which many companies respond to such outages. In the case of Apple, the blogosphere teamed with reports that some Apple stores were turning “cash-flush” customers home with nothing following the crash. And Wagner writes that Apple “compounded customer hassles through negligence and failure to anticipate demand.”

So, what should companies do when their customers are left in the cold? Rich Miller of Data Center Knowledge writes that “a good first step is to get over ‘downtime denial’ and accept that your communications strategy is often as important as your efforts to restore service.”

We agree. But we couldn’t agree more with Wagner’s assessment: “Apple needs to sit down and figure out what went wrong (and) make sure it never happens again.”

What About Hybrid Virtual-Physical Environments?

In a previous post, we introduced some of the splendor and snares associated with a virtual network and said that new management technologies were needed to deal with the complex nature of virtualization. We’ve only scratched the surface.

The larger and more complex a business, the more IT applications it will have. And just as virtualization isn’t for everyone, it’s also not for every application. Some applications and application components lend themselves to virtualization, while others do not. Plus, with IT budgets in flux, managers are deploying virtualization only for certain tasks.

Hence, a hybrid virtual-physical environment is born. That means application components running on physical servers will have to interact with those running virtually, a fact that leads to new complexities many IT departments never thought of six or seven years ago. Most organizations, in fact, that have deployed virtual technologies are working in these hybrid environments now. But many haven’t fully grasped how to manage them.

These so-called hybrid virtual-physical environments are going to be the norm for some time to come. Since the interaction complexity is amplified in hybrid environments, smart management tools will increasingly be needed to deal with challenges such as platform consistency.

Here are some things that a good virtualization network management system can do to deal with these problems:

• Automatically adapt algorithms to handle moving virtual components
• Correlate complex interdependencies of virtual environments so IT departments can dynamically optimize resource allocations
• Operate seamlessly in hybrid environments

-The Integrien Team

Why IT Best Practices Aren’t Followed

What good is it to have an IT administration best practices policy if you’re not going to follow it? That’s basically the question one frustrated reader – a self-described mobile technology “evangelist” – had for CIO magazine in a recent post.

“I am surprised that large companies can push best practices tools to their customers with vigor but refuse to embrace – or downright ignore – the same tools that will enable best practices in their own organizations,” the bummed blogger writes.

Most large companies have such practices. Sun Microsystems, for instance, had theirs published years ago in Sys Admin, a journal for UNIX and Linux administers.

Here’s a sample:

1. “It’s not done until it’s tested.” The type of personality attracted to IT careers, the journal says, are not the type who particularly like repetitive and menial quality control.
2. “If you are constantly fighting fires, find the source.” This goes to the core of what we believe at Integrien. Today’s complex IT systems need technologies that predict and pinpoint problems.
3. “Ockham’s Razor is very sharp indeed.” Medieval philosopher William of Ockham’s law of parsimony, often quoted, says the simplest of several theories is the preferable one. This applies to IT. Departments should perform the simplest of troubleshooting first, moving to more complex tests later.
4. “Communicate with users.” This is an obvious, but often overlooked Golden Rule that many IT departments don’t follow.

Why are these rules often not followed? Look to No. 2 for a prime reason. From our own experiences, we see that IT is too busy finding quick solutions of problems instead of the causes. It’s akin to the little Dutch boy plugging in the dyke. IT is filling holes instead of building a better dam. They’re pulled in so many directions, from directives from higher up and staffing issues to yes, fighting fires. There’s simply not enough time for many IT teams to consistently follow their own guidelines – guidelines many push on their own customers.

So what’s the solution? Predictive IT. That is, technology that prevents endless alerts and offers proactive, automated analysis of problems – before computers go down, not after. That seems like the simplest approach to us. William of Ockham would likely agree.

What the Amazon.com Outage Shows Us

If it can happen to Amazon.com, the world’s biggest and arguably most sophisticated online retailer, it can happen to anyone.

Amazon.com visitors were recently hit with cryptic errors when they tried to get on the site and were denied access for two hours. The outage theoretically cost the e-commerce giant $3.3 million when factoring in its yearly revenue of $4.1 billion. Of course, Amazon.com has other revenue streams. Further, some visitors, perhaps many, just came back later to make their purchases. Nonetheless, it was a costly 120 minutes. And it’s the second outage for the firm in four months.

Theories from the absurd to the pointed abound why Amazon crashed. The absurd: Some readers on various blogs have theorized that a new release of a Dungeons and Dragons game caused too many visitors on the site and caused it to crash. And the pointed:

• A hiccup in the configuration of a proxy server. (A proxy server is a computer system that forwards requests to other servers)

• A break in communications between the Web application servers and databases

• A misconfigured HTTP service on the load balancer. ( A load balancer spreads work between two or more computers)

• A content delivery network not retrieving the home page properly

While anyone can guess what caused it – as of June 19, Amazon still hasn’t released the reason why – there are things IT departments should be doing to prevent such costly outages. The No. 1 solution? Deploy an automated system that better predicts looming problems. According to a 2007 survey sponsored by Integrien, 61 percent of respondents said they manually correlate IT alerts, with some of them receiving as many as 5,000 alerts every month. With this much time being wasted – how many hours are needed to go over 5,000 alerts? – the potential for errors is high.

The study surveyed 72 IT managers from organizations bringing in at least $500 million in annual revenue. It shows something we’ve been saying for years: Too many IT departments spend too much time fighting fires and not enough time preventing them.

It’s clear here an alternative approach to IT problem resolution is greatly needed.

Want Real Innovation? Trust the Small Guys

Big companies have dominated the IT management market for as long as we can remember. But according to a post in Network World by Denise Dubie, these companies are facing increased threats from smaller companies. She says these smaller software makers “continue to make noise” about how size doesn’t matter if their products are superior.

In addition, Dubie writes in another post, IT management heavyweights BMC, CA, Hewlett-Packard and IBM are “barely making the grades with their customers” and posting C and D averages, according to a 2007 Gartner poll.

Software made by the biggest of companies has always been seen as the safe choice, as these mega corporations have the resources to throw large amounts of discounted (or even free) services to make sure their products work in the customer’s environment. But large enterprise software makers have never been known for their easy-to-use, seamless integration or innovative functionality. That has typically been in the realm of the companies that have to offer these benefits just to survive – the small guys.

The biggest firms, Dubie writes, are waking up to this fact though. She says they are focusing more and more on factors that have typically been in the small companies’ courts.

We don’t expect major changes, though. Big companies don’t tend to come up with most of the marketplace’s innovation. They buy it. They buy it because they can. And they aren’t as innovative as most startups simply because they don’t have to be. As George Clooney’s character says in the movie “Three Kings,” the most important thing in life isn’t love or respect – it’s “necessity.” Says Clooney: “People do what is most necessary to them at any given moment.”

This can be applied to war, the main topic of “Three Kings.” It can also be applied to business.

Of course big firms are innovative. They invented the computer and the iPhone, after all. But they’re not as innovative as smaller firms, according to Hewlett-Packard CTO Phil McKinney. At a recent “boot camp for entrepreneurs,” McKinney told the audience that small companies generate 14 times more patents per employee than big companies such as the one he works at. In addition, he said, 40 percent of HP’s innovations come from outside the company.

We’d like to know your thoughts. Do you think most of America’s true innovation comes from startups? Do you think the mega corporations are in trouble? Or will the largest companies simply adapt when necessity comes knocking at their doors?

Best wishes,

The Integrien team

Yes, Virtualization Saves Money But …

Virtualization is virtually everywhere on the IT landscape now.

In early June, Microsoft unveiled in-depth plans to make a push into this space with its Hyper-V technology, which will offer virtual services for applications, presentations, desktops and servers. Also in June, Symantec’s storage division, Veritas, also announced plans to offer integrated virtualization technologies. Microsoft’s offering is expected to be released in early August. Symantec’s offering, meanwhile, is scheduled to come out this fall.

With server virtualization, virtual machines are used to consolidate many physical servers into fewer servers, which in turn host the virtual machines. Basically, it transforms hardware into software.

What are the benefits of this practice? Instead of multiple machines running at partial utilizations, a company can run one machine at a higher utilization. This not only saves money on hardware but also on energy bills, space and cooling since there are fewer resources being run.

Virtualization is cheaper. But it causes all sorts of management and security migraines. The interactions between virtual machines are complex. As a result, most virtualization management technologies on the market right now still require substantial manual effort to solve performance problems. In order to be efficient, though, virtual environments require powerful tools, tools that better perform everything from handling interdependencies to proactively notifying technology managers of pending problems.

According to the Virtualization Industry Challenges Report for 2007-2008,  the capability to track and troubleshoot performance is recognized as a “near-term” issue. Some of the challenges the report highlights include:

• Harder identification and solutions for bottlenecks
• Identifying and isolating workloads placement
• Dealing with the higher bandwidth associated with virtualization

Because companies such as Microsoft and Virtual Iron are now clawing at VMware – one of the larger virtual network companies in the country – the price of hypervisors is decreasing. That means many companies are now managing multiple hypervisors, virtualization platforms that allow multiple operating systems to run on a host computer, at once. The end result? Providing solutions just for VMware simply will not cut it anymore.

Seventy percent of IT departments are pursuing server consolidation projects right now. Virtualization is slowly becoming mainstream. But once a virtual network is installed, it’s not over. Companies then need to do their homework on adopting the right security and management technologies.