Data-driven

There's been a general outcry lately about how vendor marketing organizations are abusing the cloud by force-fitting many new and existing products into the cloud computing mold.

Still, some cloud-like things actually do fit without the aid of a crow bar. A case in point is IBM's Smart Analytics Cloud.

The Smart Analytics Cloud is a solution set and reference model based on an IBM-internal Business Intelligence (BI) project code-named Blue Insight, which IBM claims to be the largest private cloud built to date. Blue insight has allowed IBM to eliminate multiple BI systems that were all performing essentially the same extract-transform-load (ETL) processes for different user groups.

It combines the resources of 100-plus separate systems within IBM such that 200,000 or so consumers of IBM's BI data now have a private cloud that acts as a centralized repository. Even better, Blue Insight does in fact fit the NIST definition of a private cloud. All Blue Insight users can, given the right permissions, get access to all data within the cloud.

What IBM wants you to know is that you too can build your own private information analytics cloud--the IBM Smart Analytics Cloud. But here's where you may stop and ponder. The solution set consists of a set of BI cloud services, and Cognos 8 BI software running on an IBM z/OS mainframe. You like the concept you say, but it's the mainframe part that may have you rubbing your chin.

So let's take a step back for a minute and put what you may see as a venerable, old beast into the cloud perspective. Please read my recent post on the VMware/Cisco/EMC consortium. I chided myself for suggesting that Vblock was in fact an open systems mainframe. OK, now I'm going to come right out and say it. A Vblock is an open systems mainframe. And, while it may be the first, it won't be the only one. Hewlett-Packard says you can build one with almost all of its parts and guidance, and OracleSun will likely announce one of its own once the EU relents. So put the z/OS in that mainframe in that perspective. It already supports thousands of Linux VMs.

What IBM has done is come up with a perfect application for a private cloud. Many large company IT departments, like IBM's, have multiple BI systems all essentially performing the same ETL function for different internal BI consumer groups. What Blue Insight does for these redundant and often expensive systems is very much like what a hypervisor does for redundant application servers--it blows them away. And because these systems can run into the hundreds of thousands if not millions of dollars, the savings can be more than substantial. The question for the mainframe skeptic: is the cost savings enough to justify learning, or perhaps re-learning z/OS?

You may take some comfort from this observation: the number of new z/OS users is on the rise. Why? They run virtual machines and have been doing so for decades. The systems integration work is done. The management applications are there. And security is miles ahead of the cloud alternatives now available. No waiting for maturity to come along, all in good time. You can get it all now.

This is not a shill piece for the z/OS mainframe even if it feels like one. I'm arguing that, if you're looking seriously at consolidated private cloud platforms, due diligence says you should not dismiss one out-of-hand that has stood up over time longer than any other single IT platform.

Client/server computing was supposed to have been the the mainframe killer. It wasn't. Now those redundant servers are stacking up on the loading docks of the recyclers. Just sayin'.

Earlier this week, EMC revealed that it has attracted longtime Intel executive Pat Gelsinger to run its storage business.

Gelsinger is set to become president and chief operating officer of EMC's Information Infrastructure Products (virtually all in EMC's product group except VMware), including the Enterprise Storage Division, RSA Information Security, Content Management and Archiving, and Ionix IT Management. His direct reports will be Frank Hauck, who now leads ESD, Mark Lewis of CMA, Art Coviello of RSA, and Jay Mastaj of Ionix.

A Wall Street Journal blog post quotes Gelsinger as ultimately wanting to be Intel's president, but that wasn't something that was going to happen anytime soon. In that the move is effective immediately, Gelsinger will likely not be a keynote speaker at the Intel Developer Forum, as originally planned.

Pat Gelsinger

(Credit: Intel)

Gelsinger is an interesting acquisition for EMC, as it diversifies its way toward encompassing more and more traditional IT infrastructure products and services, as well as IT virtualization and emerging cloud-computing models. His chip geek credentials are solid. He wrote the book on programming the 80386, and he designed the original 486 processor. At EMC, he will move to bring to market products based on a tighter integration between existing product lines, as well as those from VMware and some key partners (read Cisco Systems).

While Gelsinger is a hard-driving executive, he reveals an actively spiritual side of himself in his book "The Juggling Act: Bringing Balance to your Faith, Family, and Work."

While the Gelsinger move seems to have attracted the most media attention, there's way more to this story. As part of the executive personnel announcement, Howard Elias was promoted to president and chief operating officer of information infrastructure and cloud services at EMC.

Previously, Elias was president of EMC Global Services and EMC's Ionix IT management group. He will now be responsible for all of EMC's service groups, including those attached to the products groups. Elias also gets to champion EMC's moves into cloud services.

For me, the hidden word here is "services." IBM spawned IGS. HP bought EDS. There's a void here that I think Elias has to fill. Can Elias give EMC a services powerhouse? Can he successfully blend the more traditional IT services with the newly emerging cloud-computing models? As I see it, those are Elias' challenges.

What's going on at the top? First, Tucci believes that the current executive management structure needs to be enhanced and expended, as it progresses from $14 billion in annual revenue to a $20 billion to $25 billion IT infrastructure company capable of competing with the likes of IBM and Hewlett-Packard. Hence the division of responsibility along the line of products vs. services.

The Wall Street Journal reports that Tucci has, with this new management structure, set up something of a three-way, three-year competition to become his successor, as Tucci also announced that he plans to remain president and CEO through 2012. Now in the running, according to the Journal report, are Gelsinger, Elias, and David Goulden, EMC's current executive vice president and chief financial officer.

Tucci is certainly laying his cards on the table for Wall Street to see, and Wall Street appears to like the move, pushing EMC shares upward over the last week. His latest moves help discourage rumors that EMC is in play as a takeover candidate.

There is a big disconnect between how long people think they should be storing data and how long they actual can. One group of vendors and academics is trying to change that.

Two years ago, the Storage Networking Industry Association's Data Management Forum reported the results of a landmark study that looked at the state of long-term storage, i.e. preserving a digital object for more than 10 years. Some disturbing results jumped out.

The study suggested that we live in a digital version of the Dark Ages. I'm talking about it now because I think the messages from the study are still very relevant to both IT administrators and consumers.

A whopping 80 percent of the 276 organizations included in the study reported a need to retain electronic records for more than 50 years, so let's start there. How many of you storage administrators out there actually think you can do 50 years of electronic records retention given current technology? Without data loss? OK, so you won't be doing the same job 50 years from now, so why care? Next question: How many of you think that you can do more than three migrations of archival data from one storage media to the next without data loss? According to the study, the answer was very few of you.

Here's one for consumers: How many of you using Internet photo services sites think that your digitized images will still be there 50 years from now? You haven't thought about that, right? You and your spouse take pictures of the newborn today, you store them online, and maybe you store them at home, too. Here's a suggestion: make sure to print them and preserve the prints for as long as you can because if the enterprise-level storage administrators who have been doing digital storage for decades have little confidence in their ability to do long-term digital preservation, you shouldn't have much confidence either.

So there's a big gap here. A group of concerned vendors and academic advisers have formed the 100 Year Archive Task Force under the auspices of the Storage Networking Industry Association's Data Management Forum wants to start filling the gap. You can follow their progress or become involved yourself here.

One more result from the study still has me puzzled. Slightly more than half of the 276 organizations surveyed reported the need for "permanent" storage. What might fall into the permanent category? I thought of the Founding Fathers writing the U.S. Constitution and wondered what that process would have been like if they were all using a collaborative work-flow tool like Microsoft SharePoint. For sure, they'd print out the final version for all to see--on parchment maybe? But what about all the draft versions and messaging back and forth--in short, all the supporting documentation that clue us in on their state of mind and tell us what they really intended? Would they have printed out all of that, too? I dare say that insight would be gone forever.

We rarely, if ever, think of saving our digitized thoughts for the sake of posterity. But for the sake of historians, lawmakers, sociologists, and scientists yet to be born, we should--or people centuries from now might look back on this as the digital version of the Dark Age centuries from now.

As yet, Twitter is likely not on anyone's list of the top 10 most-critical applications. But has the U.S. government given Twitter a big push toward critical application status? This week the U.S. Department of State told Twitter it could not shut down for system maintenance because it had become a lifeline for thousands of protesters in Iran.

That should change the way IT vendors (particularly infrastructure vendors) view social-networking sites such as Twitter, Facebook, YouTube, etc.

(Credit: Twitter)

Generally speaking, social-networking sites offer no guarantees to users. You post your content, you take your chances. And, while there is no sign at the entrance that says "Caution. You are about to enter a service-level-free zone," the sign is virtually there.

To infrastructure vendors, that spells commodity play. They look at the social-networking providers as great places to earn their Web 2.0 stripes, but tough places to make money. So, they are most likely to sell them least common denominator servers and storage--no frills, no value adds. In storage, for instance, they may only be asked to supply JBOD (just a bunch of disks) storage without RAID-based data protection, snapshots, or other quality of service enhancers. But that's OK they figure, no guarantees equals no risk.

Time out guys, there is risk. Think back 10 years ago today. eBay outages were in the news on an almost daily basis and Sun Microsystems wound up wearing the blame. Yes eBay was charting territory in a brave new world and therefore offered no service or availability guarantees. And yes there were more vendors in the mix at the time (Oracle and Veritas to name two). But the outages were very visible and Sun's image suffered disproportionately. While not explicitly stated, eBay users nonetheless had an implicit expectation of quality of service from eBay, a level that was never formally agreed to, but understood and expected.

Fast forward 10 years. Twitter is in uncharted territory, too. The temporary and periodic "system busy" messages are tolerated by users, but not without complaint. Jokes about Twitter's Fail Whale are common. Hey, it's not a critical app. We're all just having fun here, right? However, the elections in Iran have changed that perception. Twitter and other social-networking sites have become windows on a pivotal event with worldwide implications. The world wants to watch. Indeed, what the State Department's request says is that the whole world needs to watch. As a infrastructure vendor in this new and uncharted environment, do you now want to be blamed for an outage? For data loss? For a security breach?

This all adds up to the Twitter Conundrum. The owners of Twitter and other social-networking sites aren't likely to buy highly available, highly secure, redundant systems and storage of the type common to 24 by 7 production data centers. Their business models simply won't support big enterprise gear. But does that stop the federal government from stepping in and saying "sorry, you can't go down right now, not even for a few hours?" No. Twitter, YouTube, and FaceBook have created windows on the world, windows that could in fact change the world for the better. You can't fail (whale).

Here's the conundrum: No one presently pays a fee for posting to these sites. You get what you pay for or, in this case, you don't get what you don't pay for. You don't pay for and therefore don't get guaranteed availability or data integrity. Is the federal government now willing to subsidize Twitter so that it can function like a production data center? Probably not. Are users willing to pay a fee to get a guaranteed level of service? Again, probably not, at least not in the near future.

Owners of the social-networking sites have managed this conundrum by rolling their own. They get cheap, or even better, free infrastructure and make it work. The power implicit in what they do with the scarcest of resources is truly awesome. Now, as they're sites become embedded in the fabric of society, can they keep that model going? Perhaps, but they will likely need our help. Remember, e-mail was once a frivolous application.