Tech Trends

For Storage Vendors, Its The Economy, Stupid

2009-12-09T14:50:00-08:00

When times were good, storage management vendors could focus on “doing more:” Managing more data, providing more performance, and faster backup and failover.

These days the mantra is, to use the cliché, “doing more with less,” so vendors are often pitching how they can wring more capacity and performance out of existing disk before customers must buy more. They’re also looking to undercut established vendors with hardware or software that does much of the high-priced brands can do, but at much lower cost.

Symantec hit all these notes with its announcement this week of new versions of its Veritas Storage Foundation, Veritas Cluster File System and Veritas Cluster Server software. The rollout focused on four areas, each of which speaks directly or indirectly to the tough economic times.

1) Support from IBM and 3Par for the Thin Reclamation API developed by Veritas (now owned by Symantec.) Thin provisioning means setting aside disk space for a volume or application, but only populating the actual disk space if and when it’s needed. This only helps, of course, if you can actually reclaim the unused disk space, which Symantec says can now be done automatically on IBM and 3Par arrays (with more arrays to come, Symantec says.)

2) Expanding its capabilities to automatically reclaim used storage when moving data from traditional “thick” provisioned arrays to thin provisioned disk. (If only 50 percent of a volume on a “thick” provisioned array has been used, only that 50 percent is used on the target “thin” array.) Symantec also added wide-area migration capabilities to its previous local-area data migration among arrays from multiple vendors. This makes it a less expensive alternative to traditional wide-area replication software such as EMC’s SRDF (Symmetrix Remote Data Facility) for customers who don’t need or can’t afford high-end Symmetrix hardware at both ends of the wire.

3) Beefing up Storage Foundation with support for Microsoft’s Hyper-V virtualization technology, with greater visibility into both legacy and thin volumes. The news here is that Microsoft’s server virtualization platform seems to be getting enough traction that vendors such as Symantec must support it, even if only to let customers hold a possible Hyper-V purchase over VMware’s head as a bargaining tool.

4) Storage-aware application clustering for near instantaneous recovery of applications by integrating the Veritas Cluster File System with Oracle, Sybase and IBM DB2 databases. This gives customers who can afford a few seconds of downtime (as opposed to no downtime) a less expensive alternative to the database vendors’ zero-downtime clustering software. Customers can get this near-instantaneous recovery, Symantec says, at about the same cost as traditional Veritas Cluster Server configurations which could take tens of minutes to recover.

Another new feature in the announcement is increased support for solid-state drives, which are fast but still too pricy for any but niche applications, as the Symantec folks admit. Once again the message is: No matter how great it is, if it’s expensive and I don’t need it right now, fuggedaboutit.

Google: The Anti-Microsoft?

2009-12-04T06:26:03-08:00

Conventional wisdom has it that Google is everything Microsoft is not: Web-centric vs. PC-centric, engineering-driven vs. marketing driven, and claiming its mission is to change the world, not just to make grubby profits (although Google will admit it’s OK, they suppose, that they made $1.65 billion in its most recent quarter.

A look beneath the surface, like I just took for InfoWorld, shows a lot more similarities than differences. Like Microsoft did with Windows, Google keeps working at products until it get them right. For example, note Google’s multi-year rollout of its cloud-based Google Apps, which began as a free ad-based service to consumers and small businesses, but I found is now robust enough for even many large companies.

Like Microsoft, Google is also building an “ecosystem” of partners who depend on it for success and enthusiastically tout its offerings. I was struck by the religious-like fervor of the Google resellers who help customers install and figure out the Google apps. They feel like they’re driving a new wave of low-cost, universally accessible cloud-based computing that will sweep away clunky old PCs like the ones you and I are using right now.

Which leads to the biggest similarity: Both Google and Microsoft are riding powerful shifts in technology.

In the late ‘70s Bill Gates understood that technical changes were shifting power from centralized server-centric computing to low-cost, high-volume distributed computing, and that whoever controlled the PC and server had a cash cow that would let them dominate the future. The result: Two decades of industry dominance, and enough money to try saving the world.

Today, Google thinks it sees a similar shift, this time back towards centralized computing (in its data centers) made available anywhere through wireless broadband to mobile devices. The Golden Goose that funds its forays into everything from mobile phones to GPS is the Google search engine, and the ad sales it makes possible.

This is where the similarities begin to fray. Microsoft is, for better or worse, at heart a software company. Critics like The Register’s Cade Metz say Google is, at heart, an advertising company, and that’s why it’s not developing a true operating system (claims to the contrary) nor providing a true development environment for independent software vendors.

In that’s true, Google can take Microsoft’s place as the industry’s Evil Empire, luring everyone onto the Web so they can read ads targeted to them by Google. The more things change…

Tide Launches iPhone Stain App; Customer Patience Levels Drop to Zero

2009-11-13T10:19:47-08:00

About 15 years ago I visited Microsoft for a briefing on their “Information At Your Fingertips” strategy. Good idea, I thought, though a bit self-evident. Where else would you want your information?

Their strategy is now a reality (though stolen by Apple and the Web) with new applications like the “Stain Brain” app for the iPhone, developed by – who else? the makers of Tide. With this application you can compare you stain with other stains on line, find out how to remove it (hmm, Tide, Tide, and more Tide, I suspect) and upload your own stain-removal ideas.

At first, I wondered why someone couldn’t just read the back of the box, or Google “stain removal.” But then I realized that we are all becoming junkies on getting the information we want, when we want it, wherever we want it, and it has to be in a fun, graphical form. And we need to be able to talk about how we erased that stain and, heaven forbid, how we got it, too.

B2B buyers are getting hooked on this instant gratification and, increasingly, want it in their professional lives too. The quickest route to booking a sales call, a white paper download or just a click-through is to give them the information they need, when they need it, in the form they need it.

That means “content marketing,” “lead nurturing,” “funnel management,” “customized content” or whatever buzzword you use to describe tailoring content and delivery to where the prospect is in the sales cycle. In addition to all the other channels you must feed, you’d better have a mobile, social strategy for sales and support as well.

The next question: How to manage this on top of all the other content monsters you must feed every day? Would love to hear some answers.

Technorati Tags: B2B marketing,content marketing,consumer marketing,lead nurturing

Now: Virtualization Value Meals

2009-11-06T09:41:00-08:00

If you’ve ever had a kid you know about product bundling. Fast food chains combine a burger, drink, fries and – most importantly – the toy in a colorful box. The customer escapes the agony of their three-year-old choosing specific items, and the chain prices the combo high enough to meet their profit objective. It works because it’s simple.

Based on early feedback, it seems like Cisco, VMware and EMC are a few fries short of a Happy Meal in pitching their bundles of physical servers, virtualization software and storage hardware to customers building private clouds (internal virtual data centers.)

Bundling can provide several benefits, but the new joint venture formed by the companies to sell these bundles has to pitch the right benefits to the right customers.

One benefit is lower costs, but most big companies can already beat up vendors on price – especially in today’s market. Simplicity is another potential draw, but large companies will still need to integrate these bundles with their legacy gear and applications. Even EMC says customers will have to do 20 to 30 percent of the required customization for their bundles.

Providing “one throat to choke” in the event of problems is a possible benefit of buying a bundle. Buying from a single source cuts down on the costs of vendor management and possibly also of troubleshooting. But that benefit, in turn, could be offset by the threat of vendor lock-in, as some of VMware’s virtualization software will be tweaked for use in these bundles.

Cisco and EMC/VMware are smart companies, and I’ll bet they’ve figured out a lot of these issues. But customers haven’t gotten the message yet. Hey, maybe they should talk about the free toy.

Ellison Keeps the SPRAC Alive?

2009-09-24T09:44:16-07:00

Here’s a story I’m surprised hasn’t gotten a lot more play: Oracle CEO Larry Ellison says he plans to not only keep the Sun’s SPARC microprocessor technology alive, but to increase investment in it (assuming Oracle’s acquisition of Sun passes European antitrust review.

SPARC processors were one half of Sun’s one-two punch (the other being its Solaris operating system) that drove its success, especially in the financial services business, before the dot-com bust. Since then, Sun has been wandering from one strategy to another, essentially giving up on promoting SPARC in favor of storage and server hardware based on Intel’s X86 architecture just like everyone else on the planet.

Does this mean Oracle is going to be selling SPARC-based notebooks? No way, says Ellison. He wants to compete with IBM as a “systems” company (read: hardware and software combined) that can take on the really big jobs like running airline reservation or banking systems. In fact, he says, his model for Oracle is IBM when it was the prime “systems” company that determined the competitive environment in which all other vendors played.

Ellison’s vision seems to be tweaking future versions of SPARC to run with Solaris and Oracle to deliver really great database, virtualization, and cloud computing platforms. But isn’t that what Sun has tried to do for years with SPARC, without notable success? Haven’t a lot of the big SPARC-based Sun boxes always run Oracle databases? Can Ellison, who grew up in the software, not the hardware business succeed where Sun couldn’t? And does SPARC actually have the oomph to differentiate future Sun/Oracle software? (Some analysts say maybe; others doubt it.)

If Ellison sticks to his word, SPARC could reemerge as a competitor to Intel’s X86 architecture – not in the commodity desktop and server space, but in the more rarified world of specialized database and storage servers and network switches. Among the targets are not only IBM itself, of course, but Cisco with its Unified Computing System push into the server market. Give Ellison credit, at least, for keeping competition alive and reaching for the next big Brass Ring instead of just hunkering down to get through the recession.

Just the Virtualization Facts, M'aam

2009-08-25T15:39:47-07:00

Among the usual flood of pre-VMWorld announcements was an unusually understandable and to-the-point briefing from VKernel. Its pitch: making it simple and inexpensive (or at least simpler and less expensive) to create more VMs on a given set of physical servers.

The basic problem, says Vice President of Product Management and Marketing Kevin Conklin, is that while VMware claims customers should be able to run 10-12 virtual machines on each physical server, most are stuck at 5-7 VMs because they can’t see well enough into the virtual infrastructure to determine which processors or physical servers have the processor, memory or storage capacity to handle more VMs. That’s where VKernel comes in, a one year old company with $7 million in funding and about 200 customers, comes in.

Its Capacity Analyzer virtual appliance creates reports (click on the image below for a larger view) that even I could understand showing the best places to create new VMs and how to eliminate or avoid bottlenecks with existing VMs. “We’re maniacally focused,” says Conklin, “on providing you with an answer, in an actionable format, so you can actually fix something with this product.”

This week VKernel introduced an Optimization Pack that features three new applets that not only show you which VMs are overloaded and which are unused and can be culled, but let you single-click to ease the processor and memory bottlenecks. While it can find wasted storage or storage-related bottlenecks, CA can’t yet go in and fix those problems. Not only does it take more work to integrate CA with storage vendors’ APIs, but most storage admins fight tooth and nail against anyone messin’ with their volumes.

Even so, VKernel claims some customers at getting instant return on investment just by using CA to find and then manually delete unused VMs, often saving (it claims) 10 to 20% of a customer’s overall storage. That's an approach even a CFO can love in the downturn – especially given VKernel’s pricing, which it claims is a third to a fifth of competing products.

Chargeback Too Hard? Try Shameback.

2009-07-29T14:48:53-07:00

Remember chargebacks? It’s the practice of tracking how much bandwidth, CPU time, disk storage or other expensive IT stuff a specific employee or business unit is using, and then charging them for what they used. The idea is that if people have to pay for what they use, they’ll use less of it.

Great idea, except it gets so complicated and cumbersome that very few IT organizations I talk to these days bother with it. Half the time, they don’t even know how many virtual machines they’re running or how much disk space they have, much less who’s using them or how the heck to charge them for it. Then there’s the monumental hassle of getting agreement on chargeback formulas, as well as all the paperwork required to issue and pay the bills. It doesn’t quite make sense, especially in a recession, when you remember you’re only talking about “funny money” – shifting pocket change from one pocket of the corporate pants into another, rather than spending your time getting more money from customers.

But how about if you could just show people how many IT resources they were wasting, holding them up to public ridicule – or at least pressure from their bosses – to hold down their ravenous appetites? That’s how some customers are using Hyper9’s Virtualization Optimization Suite, according to the company’s Director of Product Management Jonathan Reeve. It’s called “showback,” he says, when the offender is just shown the error of their ways – say, a report listing all the VMs they created to test a new application and never shut down, leaving them to eat up power, CPU cycles and disk space. “Shameback” is – well, when the conversation gets more pointed. If Sigmund Freud were a data center manager, he would have figured this out long ago.

Half a Pound of Security, and Slice It Thin

2009-07-21T10:57:48-07:00

Customers want choice, they want control and they want low-cost. Zscaler does a great job of promising all three, with a deli-counter style approach to marketing that lets customers choose what they need while touting the quality of what they get.

Zscaler delivers Web security in a software in a service (SaaS) model. Rather than installing hardware or software on site, all a customer has to do is configure their clients’ Web browsers so their traffic first runs through one of the global data centers where Zscaler runs its gateways. The gateways act as proxy servers for all traffic, monitoring for inbound threats such as viruses as well as outbound vulnerabilities such as data leakage.

Because the service was built from the ground up for SaaS, Zscaler claims each of its gateways can handle 50-100 times as many transactions per second as its competitors with far less latency. Customers can choose to buy any combination of four modules: security, management, compliance and bandwidth, according to Vice President of Security Research Michael Sutton. Among other capabilities, Zscaler allows for granular management of user activities, such as allowing them to view but not upload videos to YouTube. Pricing, depending on which combination of services a customer chooses, runs between $1-3 per user per month, says Sutton.

And from a content marketing perspective, their Web site does an unusually good job of “indexing” the content by subject area. The marketing message stands out from other “cloud computing” pitches because it is crisp, convincing and well-delivered, giving the impression they have so many smart things to say you wouldn’t have time to listen – but wish you did.

Not Files, But File Servers, In the Cloud

2009-07-02T18:25:26-07:00

If plain storage is a commodity, on-line storage is even more of a commodity. So figures Egnyte Corp., a one-year-old startup that sells not cloud storage, but cloud storage servers, which look and act just like a local file servers but without all the hassles of, like, buying and maintaining equipment.

Egnyte is also different from storage in the cloud players like Carbonite and MozyPro, says co-founder Rajesh Ram, in that it’s designed to let company employees share files with each other and outside their organization, as well as regularly updates those files, not just leave them sitting untouched in the cloud in the event of a disaster. Their secret sauce, he says, is a virtualized, multitenant architecture that allows them to support as many as 450 customers with one commodity, $2,500 server.

In keeping with its “not just storage” theme, Egnyte doesn’t even price per the amount of storage provided, but instead charges about $15 per “power user” per month for up to ten users. Customers with p to three users get about 20Gbytes of storage per user, rising to a Tbyte per user for companies with up to ten users and unlimited storage for 11 or more users. Given its business, rather than consumer-focused market, these relatively low storage limits pose no problems for its customers, and keep Egnyte’s costs down to a reasonable level, says Ram. If they can indeed hold down their infrastructure costs, it’s an interesting play which might be transferable to other types of servers such as messaging, though Ram says he’ll probably OEM such products rather than develop them internally.

Building A Cloud App? First, Don't Screw Up

2009-06-15T09:48:57-07:00

My former colleague Mark Hall is predicting IBM's most recent cloud computing initiatives will fail. Mark thinks IBM's first offering, providing Web-based servers for application and test functions, will fail because it won't beat similar offerings from Microsoft and Google on price. He predicts IBM's second service, providing cloud-based "thin clients" (with most of the application processing and data done on a remote server and fed to a rudimentary, low end client device) will fail because "There is something visceral in the minds and hearts of people about having their own PCs."

I would argue there's more than just a "gut feeling" that makes me crave my own "thick client" PC with my very own processor and hard drive with all my own stuff on it. It's experience with Web sites such as LinkedIn and Web apps such as Salesforce.com.

First, the Web isn't always there. For that matter, I can't even get a decent cell phone signal half the time, even in my own house. And I'm going to rely on an Internet connection to see my own work in progress, customer lists and to-do lists? I don't think so.

Second, the Web is too slow -- not for background applications like backup but for anything that requires a crisp response, such as looking up a long-lost customer name in a rush or dashing off a quick email. There's just too much time w-a-i-t-i-n-g for screens to refresh or for drop-down lists to populate. When I'm in crank mode I don't want my computer, or connection, slowing me down.

Third, a lot of Web apps are just too kludgy. True, that's an app dev issue and has nothing to do with the underlying architecture. But have you ever seen a Salesforce.com screen? It's filled with tiny type, multiple identical-looking boxes and odd, run-around-the-block processes for something as simple as attaching a file to an email. And how about those "reply" boxes in LinkedIn, where it takes me about half a minute to figure out where my reply goes?

Finally, remembering passwords to my different cloud apps is just too much for my age-addled brain. On my PC, if I want to launch an app, I click on it -- period. True, configuring my browser to remember all those passwords is fine, except different Web apps and browsers handle this differently, and seem to forget my different user names and passwords just when I need them. Just when I get that all working nicely, a Web app forces me to change my password for my own good. At least my PC trusts me to stay me.

If cloud apps are ever going to take off for the great unwashed like myself, they'll have to 1) reliable, which requires big build-outs of reliable, broadband wireless networks; 2) vcry fast (maybe someone should get going on advanced caching techniques for thin clients and 3) more, not less, user-friendly than their client-based counterparts (which just takes good design.) Finally, give up on the "Web-only" theology and put enough code on the desktop (a la Tweetdeck, the front end for Twitter) so I can just launch it and be on the Web sans sign-on.

Until then, you can have my thick-client PC when you drag my cold, dead, fingers...well, you know the story.