We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.

Blog: Merv Adrian Subscribe to this blog's RSS feed!

Not Pictured

Welcome to my BeyeNETWORK blog! Please join me often to share your thoughts and observations on new analytic platforms, BI and data management. I maintain a vendor-focused practice that uses primary research, briefings, case studies, events and other activities that stimulate ideas as a source for commentary on strategy and execution in the marketplace. I believe the emergence of a new class of analytic platforms, and emerging data management and advanced tools herald a next step in the maturity of information technology, and I'm excited to be present for its emergence. I hope my blog entries will stimulate ideas that will serve both the vendors creating these new solutions and the companies that will improve their business prospects as a result of applying them. Please share your thoughts and input on the topics.



June 2009 Archives

iLuminate 4.0 has just been released, and its parent firm, Illuminate, is ramping a campaign to join the "new analytic DBMS" party. With a very different architecture based on storing and indexing values in an inverted-list like model, the "correlation database" or CDBMS has been winning some traction with a powerful value proposition: shorter "time to analytics" with minimal design time, lightning fast performance, and lower cost. It will catch the eye of its avowed target market: the "business middle class" defined by character, not size. Illuminate believes mid-sized companies and departments within larger ones are an under-penetrated market, because enterprise data warehouse (EDW) projects look too big and scary. DB and BI tech-savvy analysts and especially non-savvy ones alike often have tools and marts but no larger model. But like it or not, they can be successful without it, and they will pay for the privilege.

The new analytic DBMSs build on modern, innovative structures to eliminate the basic problem of RDBMS optimization for analytics: focus on one thing and the others suffer. RDBMSs were built and optimized for decades to perform well on transaction-related tasks and manage concurrency. As IT took on "decision support," reporting, ad hoc analysis and today's panoply of business intelligence paradigms, DBMS designers and DBAs struggled to fit their square, transaction-optimized pegs into the new round, analytics-demanding holes. In an effort to optimize all possible requirements, they now spend large percentages of their time on logical and physical design, partitioning strategies, indexing, re-indexing, redesigning, aggregate building, cube building, etc. Even the analytics-targeted products require substantial care and feeding from the start, and as they scale it becomes harder. Ask a DBA how much time she spends on maintenance and how much on adding new value.

iLuminate means to change that. The core idea is increasingly heard from BI tools vendors as well as analytic DBMS players: analysis by business people is search-oriented, iterative, and ad hoc. One question leads to another, which leads to another, and as inspiration strikes, analysts must be able to veer off in unexpected directions to pursue it. Tools are evolving to include search-based paradigms and advanced visualization to support this model; the underlying databases are too. iLuminate and its associated tool iCorrelate allow rapid preliminary analysis of data to find significant directions for relating and exploring correlations across dimensions. Data is stored in "value pools" which can dramatically speed retrieval for many common analytic approaches. Loading can proceed without extensive prior design, the time from data loading to usage is impressive, and compression rivals or beats columnar systems. If you want some examples, more product detail, and an engaging hands-on discussion, read Jason Brooks' eWeek review . Not all questions are so simple, of course - complex correlated subqueries, recursive joins and other tough problems are all worth testing - and you should. Quite sizable tests can be run on hardware that costs less than $10k.  

I've known Joe Foley, CTO of Illuminate, for a decade, and the firm's predecessor, Synera, which first built the underlying technology, was a client in the 90s. That effort eventually shut down, and Joe has taken his new firm through a robust cycle of funding and infrastructure building that has served Illuminate well.  With the launch of release 4.0 and its 64-bit, multi-threaded architecture, iLuminate is ready to take on all comers in POCs - and has built its marketing and selling effort around challenging prospects to engage in one. And although Foley is one of the benchmark skeptics himself, he'll tell anyone who asks about the tests iLuminate has run and why he beats all comers on them. Go ahead - make his day. Joe happily chronicles many of the challenges in his entertaining blog, Queries From Hell. Illuminate is also happy to talk about its dozens of customers (mostly near its base in Barcelona, Spain, but beginning to spring up in North America as well.) In the next few quarters, I hope that we'll see some installations that begin to stretch the product's scalability, which still needs to be demonstrated in the field.

Illuminate is taking an intriguing approach in its route to market, and IT Market Strategy believes it will take another substantial step forward in revving its engine in the second half of 2009. I'll talk about the business model and developments in another post.

Posted June 30, 2009 5:53 PM
Permalink | No Comments |

ParAccel, another of the analytic database upstarts, has weighed in on Sun hardware with a record-shattering benchmark that its competitors have thus far avoided - the 30 TB TPC-H. It's been two years since anyone has published a 30 TB TPC-H, and only 10 of any size (all smaller) have been published in the past year. One can scoff (many do) at this venerable institution, but TPC benchmarks are a rite of passage, and a badge of engineering prowess. The ParAccel Analytic Database (PADB) has set new records, raising its profile dramatically in one fell swoop. PADB came in at 16x the price/performance of Oracle, the prior leader (and only other vendor willing to tackle the 30Tb benchmark to date.) PADB, running on Sun Opteron 2356 servers, Sun Fire™ X4540 storage servers and OpenSolaris™, was 7x faster on queries and 4.6x faster loading the data than the 2 year old Oracle result. And because of its architecture, the construction and tuning of indexes and partitioning strategies were not needed. TPC rules are specific about having product in GA within 90 days, so one can expect to see PADB version 2.0, on which the benchmark was based, out in Q3.

ParAccel has seen some skepticism in the analyst community because of its relatively small published number of customers. It claims a dozen, and half are listed on its web site. Other vendors, like Vertica and Greenplum, have been very forthcoming promoting theirs, but both have more time in the market. PADBwas released in Q4 2007 and really began its arc in 2008; Vertica has a year headstart, and Greenplum even more. Rumors have also floated about whether CTO and founder Barry Zane was leaving. I had a conversation with Barry in late June to discuss the business and the benchmarks. He was clearly excited about the benchmarks, in which he was very involved, even working on the full disclosure report personally  - "It got to be like a hobby for me," he said - and he was quite clear that he is not going anywhere.

There is suddenly a sizable number of new entrants in the analytic database space; and (disclosure) some are clients. (Not ParAccel at this writing.)  I've posted about Greenplum, Vertica and Aster. Like several of the others, ParAccel has some roots in Postgres and is a massively parallel, shared-nothing architecture; like Vertica  it uses column-based storage. Like the whole group of new players, it is routinely winning proof of concept engagements (POCs) against traditional DBMS players. 

Structurally, ParAccel's offerings are available in either proprietary or fully commodity component-based offerings. ParAccel's description is that PADB is "standard-server-based;"  the software may be purchased and run on a variety of commodity platforms. The Scalable Analytic Appliance (SAA) offering uses "enterprise-class midrange SAN components from EMC." SAA uses a gigabit ethernetinterconnect and 4 processors at each node with dedicated storage. In either case, a leader node (the Postgres-derived code is found here), coordinates the activities of the compute nodes. A hot standby node is always part of the installation, and can step in for any failing node, including the Leader node.

The EMC partnership allows ParAccel to rely on a FibreChannel-connected SAN (in a modular, midrange form factor designed to scale along with servers) for its enterprise-class features.  Availability, backup and rapid recovery, and data replication thus become easier for ParAccel to deliver - as long as EMC's CLARiiON CX4 is in the picture. PADB is able to concurrently scan the server-based direct attached storage (DAS) and the SAN.  This "blended scan," ParAccel argues, gives it the best of both worlds.

ParAccel's technical papers do a nice job describing how "Continuous Sequential Scan Rate" (CSSR), measured in megabytes of I/O per second, describes throughput from platter to server and helps demonstrate the power of the new architectures. PADB boasts a patent-pending query optimizer, notable for its ability to handle correlated subqueries (CSQs), which feature in several of the TPC-H benchmark queries, and are often a performance stumbling block. Without belaboring the techie talk here, removing columns from CSQs can have substantial impact just as it does for table scans. Retrieving relevant columns also improves CSSR substantially. Columnar storage aids data compression substantially as well, and in combination with the benefits realized from not having to use substantial amounts of space for indexing, the growth rate of installed storage relative to raw data is improved considerably.

ParAccel claims wins over competitors such as Sybase IQ, Netezza, and Vertica as well as Oracle, and touts real-world performance numbers far better than the benchmark. Of the new architecture competitors it faces, only Sybase IQ has stepped up to the TPC-H bar to date, and no doubt there will be many win and loss claims from all the vendors over the year ahead. But PADB has vaulted into contention with this announcement, and will no doubt be on more short lists - as it should. ParAccel will also begin to see more attention from its partners, including hardware players beyond EMC: Dell, Fujitsu-Siemens, Intel, AMD and others. Sun, who made a substantial contribution to the benchmark by making much of the hardware available, may be somewhat less aggressive in the wake of its acquisition by Oracle. But ParAccel says Sun is not its most installed platform; customers are running on HP, Dell and IBM hardware already. Software partners are also likely to be friendlier as the temperature rises.

ParAccel is fortunate to have completed this work with Sun, whose acquisition by Oracle will no doubt create some reallocation of resources and priorities. This is a coup for ParAccel, whose timing turns out to be impeccable. As always, prospects should insist on a POC. And as Mark Madsen of Third Nature (his blog is here) says - "always hold back some queries;" you want to see how any database performs without heroic tuning, unless you plan to keep an army of specialists around.

Posted June 26, 2009 6:17 PM
Permalink | No Comments |

It's early yet and the whole story has not been told. But the flurry of Lucidera  "news" - mostly retweets and wild speculation - needs at least a brief response. It appears, in several stories from apparently reputable news outlets quoting thus-far unnamed sources,  that LucidEra, a SaaS BI startup, is shutting its operations down. The firm apparently ran out of money - its last round of funding was US$15.6 million in Series B venture capital funding in August 2007.

I wrote favorably in this blog about LucidEra a few weeks ago, and I confess that I did not see this coming. I am not a financial analyst and did not review their books. It's a lesson to me and to us all - good stories, good technology, and even some good customers are not the whole story. I don't and won't pretend in my blog to be an authoritative source on the financial viability of anyone. Caveats all around to anyone who plans to do business with people I talk about - I neither believe nor hope that anyone takes my scribblings as sufficient ground for a purchase. Due diligence, always!

I'm concerned that some of the early commentary seems focused on whether the failure of one company calls the SaaS business model into question. In a word, No. Companies will fail - in traditional license models and in new ones. Most startups fail. That's the real world, folks. Let's take the time to see what really happened here before we all pull the Chicken Little hardhats out of the closet. The likelihood is that one of the oldest equations in business:

(cash+ revenue) - (runrate * time)  MUST NOT = 0

was violated. it's that simple, and that complicated.

Posted June 23, 2009 7:02 PM
Permalink | No Comments |

Every organization of any size has reports - usually too many, often largely unread, and typically a gold mine of information not available elsewhere without substantial investment of time and resources, even if the tools are available. Where existing reports are in place and in wide use - such as those accompanying packaged ERP, CRM and other software, or in "production reporting systems" that have been in place for years - they are an excellent source of predigested information. Monarch BI Server from Datawatch can unlock this information for a fraction of the cost and effort associated with new BI projects - and rapidly.

The Monarch product family is hardly new: the first product shipped in 1991, and today Datawatch is a $20M company with 35,000 organizations using its products - some 400,000 users. Use cases abound but they share some common themes - "don't have time, don't have money, use Excel (and sometimes other tools) but can't get the data." And yet, there it is, in plain sight - in the middle of a lot of other data, to be sure, but somewhere in that stack of paper (when I was just a lad, it was green-bar paper and we [often mis-]typed it into VisiCalc. But that's another story.) Financial reports are by far the most typical source, but there is a wealth of case studies on Datawatch's web site - and they tell a wonderful set of stories.

Make no mistake, using this information is non-trivial unless you have automation - it needs to be parsed, filtered, aggregations decoded, and maybe some transformation applied. And then it has to be exported into the format of your choice. The good news is that with nearly two decades of experience, the Monarch product crew has seen it all and decoded it by now - and has built automation and user interfaces that make it simple and quick. Sort and filter data, merge multiple data sources, calculate new fields, build automated graphs, do trending across reports that are issued on a schedule - it's all built in.

Beyond the obvious attractiveness of its immediacy, the Monarch family of products offer some other intriguing benefits. Consider the cost of an SAP license for users who only run an occasional report. Consider the cost of running that report multiple times, for multiple people, who then do goodness knows what with the data - in different ways that may conflict. How much better to run the report once and make it available via browser with standard data export mechanisms. In many ways, the Monarch family of products ought to be thought of as a utility buy - canned reports, PDFs, etc abound in most organizations storage and are woefully underused. If you're struggling with project overload, fighting budget constraints and have a good deal of data in reports already, you should investigate the Monarch product family. There are some other products out there, but Datawatch has the riches, most mature set and ought to be on your short list.

Posted June 12, 2009 5:09 PM
Permalink | 2 Comments |

It's a shame that Sand Technology isn't doing better, because the technology is apparently quite good and has been for years. Decades, indeed, but after all that time, Sand is a US$7M firm. Why is that? Their June visit to the Boulder BI Braintrust (BBBT) offered some clues. And some followup into public filings paints an unattractive financial picture that suggests any prospects should proceed with extreme caution. Sand's phenomenal compression of data for "nearline storage" and analysis sounds good, but it's hard to ignore their finances.

Sand (SNDTF on the OTC BB) reported  a loss for the year ended July 31, 2008 of $1.3M, compared to a loss of $2.5M for the same period the prior year. It's very hard to find any financial information on their web site - investors are referred to DeJong and Associates for the information. That was useful, but the "Our Take" section was very over-enthusiastic. I went to www.sedar.com to look at their public filings. The results? They have lost money for the past 5 years (albeit at a declining rate. ) Their sales have grown from $5.3M to just under $7M in the same time period (they dropped from 2005 to 2006). They've cut R&D every year. As of their last filing, they are almost out of cash.  In an alternative DBMS market where emerging firms are adding customers rapidly and getting new funding, it's hard to recommend Sand, however attractive their technology may appear to be.

Their visit to the BBBT is documented by Shawn Rogers on the BBBT blog here. Arthur Ritchie, President and CEO, is a charismatic man, with a wonderful facility for telling stories, an engaging ability to challenge thoughtful people to re-think fundamental assumptions, and a visible passion for his product. Our meeting with him was no exception. Many of us know of Sand, and a predecessor product called the Nucleus database, from the 1980s. We revisited the history. recalling how the firm started working with Hitachi decades ago, and its many years attempting to market its technology. Curt Monash has also looked at Sand, and offers a bit more historical detail here.

In that meeting, though, some of us found an absence of crisp messaging. The 5 content slides provided for the presentation didn't provide any detail on market segmentation, product naming, pricing, sales structure,  partners, and the other kinds of facts analysts want. (You can see some of those things on their web site, though.) I left early because we were rambling; I was not alone. I was remote, on the phone, and that makes it a bit easier to go, but if I were in the room, I still would have excused myself. As an analyst, one of the things you want to see is how the company presents itself. I had little clue from the materials provided. If Arthur is making the sale, it may go well - he's very engaging. But you can't scale Arthur. Perhaps that's why the sales numbers have been so flat. 

Sand's website describes them this way: "SAND Technology provides data management software and best practices for companies who have large amounts of data that they need to store, access, and analyze just-in-time, and on-demand while leveraging existing infrastructure, improving operational performance and lowering TCO." Sounds like they threw everything they could into that statement. The product portfolio includes Sand/DNA for:  Oracle, DB2, MySQL, SAP, SAS, Syslog tracker. Some of these things are not like the others. But the basics are straightforward - incredibly compressed data - keep and analyze it all.

Sand claims 100 customers in the usual industries: Retail, Financial Services, CPG, Utilities, Healthcare, Telco, and Transportation. A year ago, in a briefing for me, they said there were 70; now they claim 100, but the revenue number only went up a few hundred thousand dollars. As Monash remarked to me in a conversation about Sand, "if there are 30 new-name customers in a 70 customer company with flattish revenues, repeat business must have fallen off a cliff." The customers mentioned on the website include mostly European firms:  T-Systems. Entega, Aegon, AOL Deutschland, Chus, Deutsche Post, Dunnhumby, Klingel, O2, Physicians Mutual, Wanadoo, Siemens, nhn, RI-Solutions. Not many US firms; the firm is more established in Germany, and clearly hopes an SAP connection will prove helpful. Sand has made announcements with SAP of late, and has certified their solution with SAP.

Their site also lists partnerships with Oracle, Accenture, Sun, Cap Gemini, OpenText, EMC, BearingPoint, KXEN and Information Builders' iWay. Of these only KXEN and iWay have any mention you can find with a quick search on their web site. So there is little evidence of effective partner marketing going on.

I tend to dwell a lot on execution in my work with vendors, clients and non-clients alike. I want to know how the message is delivered, how many sales teams they have, how the comp plan works, what the partner program looks like. And some vendors who brief me are surprised that I don't want to spend more time on deeper technical details than I do. Here's why: good technology is necessary but not sufficient. The winning offerings know who their prospects are, what problems they have, and how to position themselves as solving those problems. They know how to reach those prospects, and how to close business. They know how to support them, and sustain or even increase revenue by meeting their needs. I talk to their customers and partners if I'm digging deep, and I expect to learn as much from that exercise as from technology slideshows and demos.

When I talk to a vendor who leaves most of that out, maybe it's "just bad marketing." But if the financials are not good, and momentum is not there, and the partners aren't visibly committed, that's enough to give me pause.

Posted June 9, 2009 5:49 PM
Permalink | 4 Comments |