JK2—SQL is purely to access and manipulate structured, relational, tabular data sets. But, considering that there is far more unstructured and semi-structured information out there, it’s probably best that we don’t try to force-fit SQL into all of that. Considering the range of access protocols in SOA, Web 2.0, and other services environments, it’s better to treat SQL as just one language to be supported in a broader information virtualization environment.
JK2—This is the only sustainable value-based pricing model in the DW market. Vendors will continue to assert 10x and greater performance improvements vis-à-vis the completion, or vis-à-vis their previous versions, or vis-à-vis what customers are doing now. The only way for the poor customer to know if these claims hold water is for the vendor to demonstrate this enhanced performance on the customer’s own queries and data sets. If the vendor can demonstrate an order-of-magnitude improvement, they can justifiable charge customers a premium. If they can’t, the vendor should bear the financial risk for its hype, and give the solution to the customer for free (if they’ll have it). Secondary issue: Whether the vendor should also provide free or discounted maintenance, support, and upgrades, and for how long.
JK2—These guys are one of several pure-plays with the full range of DW form factors (note: when I said “cloud” in this tweet, I meant public “SaaS,” which Kognitio calls “DaaS”). About half of Kognitio’s revenues come from DaaS customers. Considering that Kognitio has been offering this for several years, I consider them one of the most mature providers of “DW in the cloud.” In other words, they’re just a little ahead of the market—though, of course, Aster, Vertica, and a few others offer public cloud/SaaS-based DW services.
JK2—The core definition of data mining is that it identifies non-obvious patterns in historical data sets. But all of this statistical gold digging can’t serve business if it doesn’t produce a steady stream of nuggets meaningful to business people who don’t have degrees in advanced math.
JK2—I like to take refresher briefings from all DW vendors every 3-6 months. With ParAccel, they’ve benefited from the steady growth in the DW space as a whole, but also from their own aggressive pricing, marketing, and sales—plus outstanding progress on scaling their massively parallel DW appliance up and out.
JK2—Just curious: Do those online dating services use social network analysis to identify those myriad “dimensions of compatibility”? If so, do they update those models based on actual results, in terms of people hooking up—or not--based on those dimensions? Or that all pure hogwash? Preying on the desperation of lonely people?
JK2—Just curious, nothing more. Seeing as how Google has made so little headway in the enterprise market, I doubt they could make much of a go with a BI SaaS service. I do expect Microsoft to integrate Bing technology into their BI stack in the next 2-3 years. Calling Bing a “decision engine,” and giving it a name that begins with “B-I” sort of hints at that direction.
JK2—Solid state drives offer order-of-magnitude performance, reliability, and power-consumption advantages over traditional storage media. No one seriously doubts that rotating media will disappear over the next decade or so. Cost parity (per TB) will be achieved by the middle of the ‘10s.
JK2—SSDs will become the preferred storage technology in DW appliances by 2015, I predict.
JK2—As the list price of DW appliances approaches $2,000/TB (one-tenth of today’s leading-edge $20,000/TB), the average size of enterprise DWs will grow by a factor of ten: into the 100s of TBs. That will be in the middle of the coming decade. As the list price of DW appliances drops by yet another order of magnitude (to $200/TB), the average size of enterprise DWs will reach the petabyte level. That will be the end of the coming decade.
JK2—In 10 years, you will be as likely to find a rotating disk on a new DW appliance as you are to find a tape drive on today’s appliances.
JK2—Think of all the unstructured and semi-structured content streaming in social networks. Think of all the information in your enterprise’s content management systems. Think of all the compliance-relevant data in your e-mail systems. Think of all the audit logs in all of your transactional systems. Think of all the years worth of historical information you will want to persist from all applications. Think of all the streaming event data you will need to aggregate to drive complex event processing applications. Annual doubling? That’ll sound quaint when your tsunami demands annual quadrupling and beyond.
JK2—The only word in the bunch that Beavis and Butt-Head would have snickered at is “dongle.”
JK2—Can someone point to a specific business opportunity that Google refused on the grounds that it would have required them to undertake a joint venture with Lucifer?
JK2—More than three-quarters of today’s DWs are in this range, by my rough reckoning.
JK2—And, quite frankly, search engines strike me as deadly dull. Albeit indispensable.
JK2—Without a strong open-source push, it’s not at all clear that Google Chrome can ever gain broad adoption. Linux is there, and it’s done it in a decade or so.
JK2—I’m a bit jaded about event processing becoming the next big thing. It’s big, all right, and it coming to the BI space in a big way, but it’s not a new thing, and not something that will burst into universal adoption anytime soon. Some analysts have been claiming for several years that “event driven architecture” is the next big thing.
JK2—Event processing can be the next big thing in specific application domains. Event data warehousing, for example, which could provide log aggregation for security, e-discovery, and predictive analytics.
JK2—In Q1, I’ll be developing a Forrester report on virtualizing the DW in to the cloud, and evolving it into a complex content warehouse for advanced analytics.
JK2—Social network analysis is all about mining the statistical patterns in people’s behavior. It’s also about mining the substance of the things they say—to each other, to each other, and to no one in particular. What’s on their minds. And how their thoughts surface aspects of a collective social intelligence.
JK2—Bring down the cost of storage, and the average EDW will grow accordingly. Pent-up demand.
JK2—Public cloud storage is where petabytes will enter the enterprise DW picture. Through outsourcing of staging, archiving, and backup to cost-effective, scalable clouds.
JK2—Remember that Moore’s Law is just a trending rule of thumb. It’s not actually written into the immutable laws of the universe.
OLAP is all about fast, flexible, multidimensional queries against large, prejoined, structured data sets. Not great cocktail party talk. 9:27 PM Dec 6th from TweetDeck
JK2—Social networking isn’t great cocktail party talk either. It’s become so common that it’s now like discussing who’s got the coolest e-mail signature. Best experienced, not discussed.
JK2—One big difference: Using OLAP won’t get you suspended from Major League Baseball.
CONTINUING-TO-INDULGE-IN-SOCIAL-MEDIA-ARE-CHANGING-THE-VERY-FABRIC-OF-OUR-POSTMODERN-EXISTENCE-WAIT-A-SEC-WHY-AREN’T THESE IN TECH-TWEET? TWEETS
STRAY PERSONAL TWEETS
STRAY POLITICAL TWEETS
STRAY SCIENCE TWEETS