Virtualization

Pervasive BI scenarios, from search technology to the cloud

Whenever I go to a show related to business intelligence or speak with an expert in the field of BI, there are endless opinions on how to get more out of BI and get that data out to more users. One answer that keeps coming up is search technology; another is cloud computing.

Back in June at BI vendor Information Builders’ user show, many attendees spoke of the benefits of marrying search technology and business intelligence. At the time, the Royal Bank of Canada was beginning to look at search technology to mine unstructured data in dead repositories. The 70,000-plus-employee company has unstructured data all over its enterprise, but the main target for now is archived enterprise content management repositories.

Also at the time, Boris Evelson said the future of BI lies in appliances that crawl both structured and unstructured data and places all information in optimized storage, where it becomes instantly available. The front end requires no IT support and users can explore the BI environment using search-like capabilities with no limitations, whether data is in email or desktop docs — making BI at last pervasive throughout the enterprise.

In a more recent conversation with SaaS BI vendor Birst’s CEO, Brad Peters, I heard a bit more skepticism on the topic of business intelligence applications being replaced by search technology. “It seems that every six to seven years that becomes a hot topic and you start to see BI companies trying to classify themselves as search companies, and it has come around once again,” he said.

It is true that BI tools are looking more like search, giving users the ability to pop up search-like dashboards and interfaces, but turning a search technology into an outright business intelligence tool is not that simple.

“The reason BI [tools] are successful is because business rules and the business logic around how your business is run is encapsulated in the BI solution, and that is created by adding human capital. And that’s something you can’t really pull out of search,” he said.

If you do a search on company revenue, for example, the result could be calculated 15 different ways based on what the user punches in. So can you trust that data?

“How was that data calculated? Where did it come from? If the answer is ‘I don’t know,’ then you can’t trust that number, and if you can’t trust that number you don’t have BI,” he said.

The idea of giving users a search box so that any user, with any level of skill set, will finally have access to BI, does appeal to Peters.

Yet, some argue that BI is not meant for the masses, but the business analysts within the company whose sole job is to crunch the numbers based on criteria set by executives.

As for the future of BI, Peters sees it in the cloud. A place where all your data can be accessed by logically integrating data, versus physically, giving data access to anyone in your organization.

Yet, when you bring up the marriage of search and BI to some experts they cringe, arguing that BI alone is hard to manage, never mind introducing another technology into the mix or adding unstructured data such as email or word documents into your BI strategy.

The debate continues, but it is clear that BI and search technologies can learn a thing or two from each other.

Give us your take in the comments.

With PPM software, don’t always throw in the kitchen sink

Project and portfolio management (PPM) software is definitely not overlooked in the midmarket. According to a recent SearchCIO-Midmarket.com survey, 17% use some form of PPM software. But how many of those organizations are actually using all of what they paid for?

Time and time again, I hear the stories of the initial implementation process: Company purchases software, company implements relevant parts of the software in stages, company stops the process once it has implemented what it needs.

I once bought a 160-piece kitchen set. Out of the 160 “useful” kitchen tools, I use about 10 of them — the rest sit in my cabinets, waiting for the day I decide to make eight individual servings of crème brûlée or require a tool for removing my strawberry stem. At the time, I had big plans for my cooking skills and thought the new tools would inspire me to learn a bit more in the kitchen — maybe I really will try my hand at beef Wellington, I thought.

Nope.
Continued »

Five key questions about cloud computing

It seems every time I ask someone what they think about cloud computing, I’m asked five or more questions in return:

Isn’t what cloud providers offer pretty cookie cutter? Translate that as, “They might not support the particular platforms and configurations I have.”

What about licensing? Just the other day, someone at a business intelligence show said it wasn’t clear how database licensing would be handled. Would he license the database, or would the cloud computing provider?

Which led to a question many people have: If I host an application, such as a business intelligence application with a cloud provider, it would have to connect to my data sources. Wouldn’t that be a security risk?

How will that impact compliance with regulations?

And the most popular concern: Why would I want to expose my data sources like that? Continued »

Can startups get to market quicker via Amazon’s cloud computing model?

Amazon’s cloud computing model is allowing startup companies like Confidela to bring their products to market quicker and cheaper.

The company of 15, which recently launched WatchDox, a document and permissions product, decided to utilize Amazon’s Elastic Compute cloud infrastructure (EC2) rather than to invest time and money in a new data center.

“For a startup just launching, we get huge data center capabilities,” said Confidela CEO and co-founder Moti Rafalin. “It’s one of the benefits of this era.”

Previously, a small startup company like Confidela would need to invest tens of thousands of dollars into a data center. Now, Confidela and other small businesses can take advantage of the scalability and “buy as you grow” option available with public clouds like Amazon’s – a natural way to grow and progress over time. Continued »

Update on move to Gmail: Cost-cutting measure a mix of sun and clouds

I checked in with early cloud adopter Jerry Hodge, CIO of Hamilton Beach Brands Inc., for an update on his pioneering migration from Lotus Notes to Gmail — the quintessential cloud app. The company, which started the project in January, is among the first ever to move to Gmail from Lotus, according to Hodge.

The backstory: Hamilton Beach was facing a mandatory upgrade of Lotus Notes that would have required not only the expense of the software upgrade, but also additional hardware and considerable staff effort. Hodge realized he could save about $500,000 in capital and operating costs over five years, and another $400,000 in labor if he went with the enterprise edition of Gmail.

In a year when his capital budget request was cut 60%, saving a million bucks seemed like a great idea, despite some trepidation on the part of his staff. He made the big switch the right way– gradually. He lined up a test group in the company’s Mexico City and China offices to try it first to work out the kinks, and he smartly waited until everything was working before the C-suite got it. People’s email would be maintained as @hamiltonbeach.com, which was important to the company.

So how did it go? Continued »

Open source cloud computing: The future midmarket sweet spot?

What happens when open source and cloud computing collide? Cost savings, flexibility and (at least one open source vendor hopes) midmarket CIOs checking it out.

Open source CRM provider SugarCRM has launched an on-demand version of its software that is included free with an on-premise license.

Subscribers can switch back and forth between the local server and the cloud version, called Open Cloud. They can make one version a hot backup for disaster recovery, or use one version for testing and the other in production if they like. “There’s no one asking you if you want on-demand or on-site; you get both,” said Martin Schneider, director of product marketing at SugarCRM Inc.

Jay Lyman, an open source analyst at The 451 Group, said the number of startups emerging pairing open source with cloud computing is a clear indication of a growing trend. “Open source is a good fit for cloud computing because of the interoperability and the portability,” he said. “We’re going to see rapid experimentation, testing and vendors using this as an opportunity to learn customer pain points and match the right apps in with the clouds.”

In many cases, small and midsized businesses will investigate cloud computing the same way they checked out open source – by experimenting with and investigating minimal fees. “[CIOs] can’t revamp their entire systems in the down economy, but they can look into trying new things while still leveraging their existing applications with open source cloud offerings,” Lyman said.

Is cloud computing a platform or a stage?

Cloud computing is at that early stage where there’s much discussion of taxonomy: Is it infrastructure? Is it Software as a Service? Is it a platform? All well and good, but little discussed so far is “Whatever it is, where does it fit in the lifecycle of corporate IT?”

Talking to IT folks who attended last night’s MIT Innovation Series forum on cloud computing, I got much more of a sense that some see cloud computing more as a staging area for new projects or businesses than as a permanent platform. Startup companies, which spend most of their first years as small to midmarket firms, are being touted as one of the early markets for cloud computing. Speaking to one startup technology director, his view of cloud computing was instructive. He is trying to walk between a rock and a hard place: provision too few servers and his company’s customers will have a bad experience online and never come back; provision too many servers and the company will run out of funding too soon.

His interest in cloud was as a prototype environment, therefore. He can find the right amount of capacity for a given number of users of his website without heavy capital investment. Later, he is likely to bring the operation in-house. He’s more impressed by the flexibility of cloud computing then by the cost: He says cloud capacity costs about 30% more than owning it, but that’s just comparing raw capacity costs, not including labor or real estate.

Still, it was evident at the forum that a lot of midmarket service providers in the technology space — development tools and middleware companies, contract developers and integrators and Web design firms, to name a few — are likely to convert their offerings to cloud offerings. Instead of buying a boutique tool vendor’s product, you’ll use it in the cloud as part of a whole stack. Hundreds of companies, literally, are lining up to offer such services on top of Amazon, and ultimately on top of other providers as they emerge. Their goal is to get you to stay in the cloud, using (and paying for) their portion of the stack indefinitely.

Hyper-V update: Yes, it works with Linux

The Hyper-V experiment continues. My three Server 2008 VMs are extremely stable — no crashes at all, nine days continuous running. I made a number of restarts to address various upgrades and all three VMs came back automatically. Am also using a VM to test Windows 7, and no problems there either.

Now addressing how well Hyper-V handles Linux. Ubuntu 7.1 desktop installed with no problems at all, although still working on networking. My Linux skills are pitiful, so I draw no conclusions yet about whether the networking issues are related to Hyper-V or user error. Perhaps I’ll have to install the Unix services roll on the base server. Am also installing a more recent version, 8.1, and will probably throw a Linux server up there, too. From a CPU and disk perspective, no problem — this machine still has a gear or two on the upside.

I am running out of memory, however, now that I have 6 VMs on this 8GB machine. The culprit: SQL Server on the base machine. Why did I install that? I’ve been trying to get System Center Essentials (SCE) on to the base machine and that requires SQL Server. This is a very tangled web Microsoft has woven: you need SQL Server and Windows Server Update Services to run SCE. SQL Server, no big deal; WSUS, something of a project, as it involves Group Policy. So far I haven’t gotten SCE to install and may abandon the whole project.

It does point out an interesting conundrum: Hyper-V manager manages memory for VMs. Is it more efficient than the base OS is? In other words, if SQL Server were running in a VM, would it be using as much memory? Sounds like my next experiment.

Should midmarket companies have one virtualization environment?

A lot of budget-strapped CIOs are going to be telling their systems and storage directors to take another look at consolidation this year. These days, consolidation means virtualization. Only recently, that mainly meant VMware ESX. That is still the weapon of choice for many reasons, but suddenly Microsoft actually has a competitive product.

After a pretty feeble offering with Virtual Server 2005, Microsoft went the hypervisor route and now offers that as a built-in feature (excuse me, a “role”) on Server 2008 Enterprise Edition. Make that your base install and you can then put any version of Windows and some versions of Linux in VMs on the same box. I’ve been playing with it on a quad-core AMD box with 8 gigs of memory and hey, it actually works! Two years ago, Hyper-V vs. ESX was a silly conversation about marketing. Now you can actually start to compare them and make decisions about how to use them.

In my case, I put three guests (a domain controller, a file server and one just idling while I figure out System Center Essentials) on Hyper-V without the box breaking a sweat. More to the point, I didn’t break a sweat, either. Even a non-MCSE guy like me could do it. No muss, no fuss. If you have admins who can install and configure Windows Server, they can work this.

There are a lot of holes in the Hyper-V story. As of 2009, it’s not going to get you close to a fully dynamic data center. You can’t move VMs around willy-nilly. There aren’t the same kind of admin tools for DR or test/dev labs or many other of the niceties that VMware and many third parties now have.

Pricewise, it might not be that big a bargain, either. Enterprise Edition can run you as much as $3,999, which isn’t very different than buying VMware VI3 and one copy of Server 2008 Standard. The devil will be in the details of your volume purchase agreements as far as that goes – depending on the support agreements, VMware could actually cost less. Eric Seibert on Server Virtualization Blog recently remarked that the many differences between the products makes comparing them, especially from a cost point of view, an apples-to-carrots comparison.

As for performance, I haven’t seen any face-offs yet between Hyper-V and Server 2008. But if you’re trying to quickly collapse a lot of low-effort servers, maybe you don’t care about the ultimate in benchmark scores.

So it comes back to what it often does when choosing between Windows and something else: familiarity, integration and ease of use. In midmarket companies, you can’t always afford overspecialized IT staff. Maybe you don’t have budget or headcount for VMware specialists. Maybe you’d rather use your existing ESX licenses for more hard-core uses like email and ERP. Maybe you want to use similar tools to manage your physical and virtual servers.

The point is, CIOs will want to take a close look at the tradeoff between having one virtual environment (whether that’s Microsoft or VMware) or tiering their virtual environments. And finally, they have a reason to do that.

Microsoft’s Midori: Straight up or on the rocks?

Microsoft is planning for the retirement of Windows by working on a new operating-system—Midori.

Maybe.

Well, more than likely considering all the Internet-buzz about it (one of many projects in ‘incubation,’ according to Microsoft).

But, Microsoft is staying tight-lipped and has not officially released any details about the code-named Midori project. According to the BBC, Software Development Times published Midori details after gaining access to some of Microsoft’s internal documents.

Midori is believed to be an Internet-based operating-system as opposed to its hard-drive-installed older brother, Windows. Rather than being installed onto individual machines, the Midori operating system would be similar to a “software-plus-services” approach or “cloud computing.” As the SDTimes put it, the Midori documents show “applications running across a multitude of topologies, ranging from client-server and multi-tier deployments to peer-to-peer at the edge, and in the cloud data center.”

Microsoft may be spot-on with this. Windows was initially popularized pre-Internet — a time when people relied solely on their very large and very stationary PCs. Times have changed. Computers are smaller and easily portable, connecting to anyone, anywhere. If Microsoft can make Midori a (properly functioning) reality, then we may be seeing a much-needed change.

Virtualization is becoming more and more popular and Microsoft will have to do something to keep up and stay on top (every PC purchased may not be pre-installed with Windows some day).

They may have started by building some hype…

It appears someone spilled the Midori, Mr. Gates.