Data Lakes, the renamed ODS, aren’t the only solution for accessing data. Think actual need, understand supporting metadata, then build your data ingestion plan. Read my latest TechTarget column.
A presentation last week, hosted by Database Trends and Applications (DBTA), was a great example of some interesting technical information presented poorly. As that sentence implies, this column is one about the marketing of business intelligence (BI), not about the technology – well, not much…
There were three presenters: Brian Bulkowski, CTO and Co-founder, Aerospike; Kevin Petrie, Senior Director and Technology Evangelist, Attunity; Reiner Kappenberger, Global Product Management, HPE Security – Data Security.
Brian was first at the podium. Aerospike is a company providing what they claim is a very high speed, scalable database, proudly advertising “NoSQL!” The problem they have is that they are one of many companies still confused about the difference between databases and SQL. A database is not the access method. What they’re really focused on in loosely structured data, the same way Hadoop and other newer databases are aimed. That doesn’t obviate the need to communicate via SQL.
He also said that the operational in-memory market is “owned by NoSQL.” However, there were no numbers. Standard RDBMS’s, columnar and NoSQL databases all are providing in-memory storage and processing. In fact, Information Management has a slide show of Gartner’s database analytics vendor report and you can see the breadth there. In addition, what I constantly hear (not statistically significant either…) is that Hadoop and other loosely-structured databases are still primarily for batch. However, as the slide show I just mentioned is in alphabetical order, and Aerospike is the first one you’ll see. Note again that I’m pointing out flaws in the marketing message, not the products. They could have a great in-memory solution, but that’s doesn’t mean NoSQL is the only NoSQL option.
The final key marketing issue is that he kept misusing “transactional.” He continued to talk about RDMS’s as transactional systems even while he talked about the power of Aerospike for better handling the transactions. In the later portion of his presentation, he was trying to say that RDBMS’s still had a place, but he was using the wrong term.
Attunity’s Kevin Petrie was second and his focus was on Attunity Replicate. The team of Aerospike and Attunity again shows the market isn’t yet mature enough to have ETL and databases come smoothly together. Kevin talked about their 35 sources and it seem that they are the front end in the marketing paring of the two companies. If you really need heterogeneous data sources and large database manipulation, you’ll need to look at the pair of companies.
My key issue with this section was one of enterprise priorities. Perhaps the one big, anonymous reference they both discussed drove the webinar, but it shouldn’t have owned the message. Mr. Petrie spent almost all his time talking about Hadoop, MongoDB and Kafka. Those are still bleeding edge tools while enterprise adoption requires a focus on integrating with standard and existing sources. Only at the end, his third anonymous case, did Kevin have a slide that mentioned RDBMS sources. If he wants to keep talking with people running experimental and leading edge tests of systems, that priority makes sense. If he wishes to talk to the larger enterprise market, he needs to turn things around.
The other issue was a slide that equated RDBMS, Data Warehouse and Hadoop as being on equal footing. There he shows a lack of business knowledge. The EDW, as an old TV would declare, is the one of these things that is not like the other. It has a very different purpose from the two database technologies and isn’t technology dependent.
Reiner Kappenberger gave a great presentation but it didn’t belong. It seems the smaller two firms were happy to get HP to help with the financing but they didn’t think about staying on message.
Let me make it very clear: Security is of critical importance. What Mr. Kappenberger had to say was very important for people to hear. However, it didn’t belong in this webinar. The topic didn’t fit and working to stuff three presenters into forty minutes is always tough. Another presentation where all three talked about how they work to ensure that the large volumes of data can be secure at multiple levels would have been great to hear – and I hope the three choose to create such a webinar.
This was two different webinars stuffed into one, blurring the message. In addition, Aerospike and Affinity either need to make sure they they’re not yet trying to address the mass market or they need to learn how to stop speaking to each other and other leading edge people and begin to better address the wider enterprise market.
The unnamed reference seemed to be a company that needed help with credit card transactions and fraud detection, and all three companies worked to provide a full solution. However, from a marketing standpoint I don’t think they did proper service to their project by this webinar.
A recent DBTA webinar was on how the data warehouse is still with us. It was by Sarah Maston, Developer Advocate, IBM Cloud Services. Simply put, it was a pitch for IBM and how their data warehousing solutions can help people more easily move to the cloud. Sarah was very knowledgeable, but she’s one of the smart folks I do suggest gets a class in presentation skills. IBM must have them and it would help her be even more powerful in her talks.
The core of the presentation was talking about how dashDB, IBM’s columnar, MPP database is perfect for data warehousing and how you can easily move information to it. Being at IBM, she had no hesitation talking about the big, visible name in Cloud: Amazon. Her claim is that IBM Cloudant is a much more powerful and agile tool for loading dashDB than is Amazon DynamoDB for Amazon Redshift. From my decades of high tech, I can believe it. IBM’s challenge is going to be whether or not they can communicate to the SMB market in ways they want to hear. That’s been a regular challenge for IBM.
One of the most interesting things Ms. Maston discussed was how to get information from systems into the data warehouse. A she said, in reference to IBM Bluemix, “meet the ODS.” I’ve previously said similar things and think it’s important to not forget the importance of the operational data store.
Data warehousing is not going away, it’s evolving. So too is the ODS. IBM is a company that often looks ahead very clearly but then sometimes misses the messaging. From the presentation, I see all the pieces are there, it’s early and they’ll grow, but it remains to be seen if they’ll learn how to address the market properly to get a major chunk of the business at which they’re aiming.
Looker’s technology is an application server that sits above relational databases to provide faster, more complex queries. They’ve developed their own language, LookML to help with that. That’s no surprise, as Lloyd is a self-described language guy.
It’s also no surprise that the demos, driven by both Lloyd and Zach, were very coding heavy. Part of the reason that very technical focus exists is, as Mr. Tabb stated, that Looker thinks there are two groups of users: Coders who build models and business managers who use the information. There is no room in that model for the business analyst, the person who understands who to communicate a complex business need to the coders and how to help the coders deliver something that is accessible to and understandable by the information consumer.
How the bifurcation was played out in the demonstration was through an almost exclusive focus on code, code and more code, with a brief display of some visualization technology. The former was very good while the later wasn’t bad but, to fit with their mainly technology focus, had complex visualizations without good enough legends – they were visualizations that would be understood by technical people but need to be better explained for the business audience they claim to address.
As an early stage company, that’s ok. The business intelligence (BI) market is still young and very fragmented. You can get different groups in large companies using different BI tools. While Looker talks about 300 customers, as with most companies of their size it could only be those small groups. If they’re going to grow past those groups, they need to focus a bit more in how to better bridge technology and business.
They also have a good start in attracting the larger market because they support both cloud and on-premises systems. The former market is growing while the later isn’t going away. Providing the ability for their server to run either place will address the needs of companies on either side of the divide.
RDMS ≠ SQL
One key to their system is they don’t move data. It stays resident on the source systems. Those could be operational systems, data warehouses, an ODS or whatever. What they must have is SQL. When asked about Hadoop and other schema-on-write systems, the Looker team stated they are an RDMS based application but they’ll work on anything with SQL access. I have no problem with the technology, but they need to be very clear about the split.
SQL came from the relational world, but as they pointed out in an aside, it isn’t limited to that. They should drop the RDMS message and focus on SQL. As Lloyd Tabb said, “SQL is the right abstraction.” What I don’t know if he understands, being focused on technology and having those biases, is it isn’t the right abstraction because of some technical advantage but because it’s the major player. McDonalds isn’t the best burger because it has the most stores. SQL might not be the best access method, but it’s the one business knows and so it’s the one the newer database companies and structures can’t ignore.
Last year, the BBBT heard from multiple companies including Actian and EXASOL, companies focused on providing SQL access to Hadoop. That’s as important as what Looker is doing. The company that manages to do both well with jump ahead of the pack.
Looker is a good, young company with some technical advantages that can greatly improve the performance of SQL queries to business databases and provides a basic BI front end to display the results. I’m not sure they have the resources to focus on both, and I think the former have the clearest advantage in the marketplace. Unless they have more funding and a strong management team that can begin to better understand the business side of the market, they will have problems addressing the visualization side of BI. They need to keep improving their engine, spread it to access more data sources, and partner with visualization companies to provide the front end.
Today the BBBT held a special session. While most presentations are by companies with full products, existing sales and who typically have been around for a few years, today we had the pleasure of listening to Sherry Brown, President of AptiMap. This is a pure startup company, still tiny. She was looking for our always vocal analyst community’s opinions on her initial aim and direction. Not to surprise anyone who knows the BBBT, we gave that at full bore.
Ms. Brown’s goal is to provide a far easier way of mapping fields between source and target datasets for creating data warehouses and other data stores. It’s a great start and she has some initial features that will help. I’ll be blunt: I’m intentionally not going to say a lot. As mentioned, they are a very early startup and the software isn’t full fledged. That means any mention of what they have and don’t have could be inaccurate by next week. That’s not a bad thing, it’s what happens at that phase.
I will mention that the product is cloud based from the start.
The important question about whether or not to contact AptiMap is what who you are and what you need. Most of the feedback to Sherry was about that. It was helping to focus the message. If I have correctly understood the consensus of the attendees, here are the critical things to focus upon while defining a market for the initial product:
- Aimed at IT and business analysts
- Folks currently using modeling tools or spreadsheets at a start
- Focus on standard, enterprise data sources, from spreadsheets to RDBMS’s, Hadoop can wait
- Mid-sized companies integrating their first sets of systems or trying to get a handle on their existing data
- Might especially be good in the hands of consultants going into those types of companies.
- Many of the potential users are tablet users, so focus on that aspect of mobile
One final key, one that needs to be a full paragraph rather than a bullet and one that many technical startups don’t get while building their products based on user needs, is that users aren’t the only decision makers in the product. As mentioned, this is a cloud product and AptiMap will be expecting recurring revenue from monthly or annual fees. The business analyst is often not the person who approves those types of costs. The firm also needs to focus messages on the buyers, whether IT, line or consulting management, to build messages that help them understand the business benefit of providing the tool to their people.
Understanding your market matters. It will help the firm not only focus product, but also narrow down the marketing message and image to aim at the correct audience.
Too often, founders get a great technical idea and focus on a couple of users to fill out product features and then try to find a market. BI is moving too fast for that, the vision needs to be much more clearly set out much earlier than was needed in software companies twenty years ago.
Finally, I mentioned the cloud model but should also mention AptiMap is offering a 30-day free trial.
AptiMap has an initial product that can help people more rapidly and accurately create mappings between data sources and targets. It’s cloud based for easy access. It is, however, very early in the product and company life cycle.
I would suggest it primarily to analysts in mid-sized organizations or consultants who work with SMBs and want some quick hit functionality add to map data sources for the creation of data warehouses, ODS’s and other relationally oriented data repositories.
If you want to experiment inexpensively with an early product that could help, contact them.
Data virtualization. What is it? A few companies have picked up the term and run with it, including last week’s BBBT presenter Denodo. The presentation team was Suresh Chandrasekaran, Sr. VP, North America, Paul Moxon, Sr. Director, Product Management & Solution Architecture, and Pablo Alvarez, Sales Engineer. Still, what I’ve not seen is a clear definition of the phrase. The Denodo team did a good job describing their successes and some features that help that, but they do avoiding a clear definition.
The companies doing data virtualization are working to create a virtual data structure where the logical definitions link back to disparate live systems instead of overlaying a single aggregated database of information. It’s the concept of a federated data warehouse from the 1990s, extended past the warehouse and now more functional because of technology improvements.
Data virtualization (and note that, sadly, I don’t create an acronym because DV is also data visualization and who needs the confusion. So more typing…) is sometimes thought of as a way to avoid data warehouses by people who hear about it at a high level, but as the Denodo team repeatedly pointed out, that’s not the case. Virtualization can simplify and speed some types of analysis, but the need for aggregated data stores isn’t going away.
The biggest problem with virtualization for everything is operational systems not being able to handle the performance hits of lots of queries. A second is that operational systems don’t typically track historical information needed for business analysis. Another is that very static data in multiple systems that’s accessed frequently can create an unnecessary load on today’s busier and busier networks. Consolidating information can simplify and speed access. Another is that change management becomes a major issue, with changes to one small system potentially causing changes to many systems and reports. There are others, but they in no way undermine the value that is virtualization.
As Pablo Alvarez discussed, virtualization and a warehouse can work well together to help companies blend data of different latencies, with virtualization bringing in dynamic data to mesh with historic and dimensional information to provide the big picture.
Denodo seems to have a very good product for virtualization. However, as I keep pointing out when listening to the smaller companies, they haven’t yet meshed their high level ideas about virtualization and their products into a clear message. The supposed marketechture slide presented by Suresh Chandrasekaran was very technical, not strategic. Where he really made a point was in discussing what makes a Denodo pitch successful.
Mr. Chandrasekaran states that pure business intelligence (BI) sales are a weak pitch for data virtualization and that a broader data need is where the value is seen by IT. That makes absolute sense as the blend between BI and real-time is just starting and BI tends to look at longer latency data. It’s the firms that are accessing a lot of disparate systems for all types of productivity and business analysis past the focus on BI who want to get to those disparate systems as easily as possible. That’s Denodo’s sweet spot.
While their high level message isn’t yet clarified or meshed with markets and products, their product marketing seems to be right on track. They’ve created a very nicely scaled product
Denodo Express is free version of their platform. Paul Moxon stated that it’s fully functional, but it can’t be clustered, has a limitation of result set size and can’t access certain data sources. However, it’s a great way for prospects to look at the functionality of the product and to build a proof-of-concept. The other great idea is that Denodo gives Express users a fixed time pricing offer for enterprise licensing. While not providing numbers, Suresh stated that the offer was working well as an incentive for the freeware to not be shelfware, for prospects to test and move down the sales funnel. To be blunt, I think that’s a great model.
One area they know is a weakness is in services, both professional services and support. That’s always an issue with a rapidly growing company and it’s good to see Denodo acknowledge that and talk about how they’re working to mitigate issues. The team said there are plans to expand their capital base next year, and I’d expect a chunk of that investment to go towards this area.
The final thing I’ll note specifically about Denodo’s presentation is their customer slides. That section had success stories presented by the customers, their own views. That was a strong way to show customer buy in but a weak way to show clear value. Each slide was very different, many were overly complex and most didn’t clearly show the value they achieved. It’s nice, but customer stories need to be better formalized.
Data Virtualization as a Market
As pointed out above, in the description of virtualization, it’s a very valuable tool. The market question is simple: Is that enough? There have been plenty of tools that eventually became part of a larger market or a feature in a larger product offering. What about data virtualization?
As the Denodo team seems to admit, data virtualization isn’t a market that can stand on its own. It must integrate with other data access, storage and provisioning systems to provide a whole to companies looking to better understand and manage their businesses. When there’s a new point solution, a tool, partnerships always work well early in the market. Denodo is doing a good job with partners to provide a robust solution to companies; but at some point bigger players don’t want to partner but to provide a complete solution.
That means data virtualization companies are going to need to spread into other areas or be acquired. Suresh Chandrasekaran thinks that data virtualization is now at the tipping point of acceptance. In my book, given how fast the software industry, in general, and data infrastructure markets, in particular, grow and evolve, that leaves a few years of very focused growth before the serious acquisitions happen – though I wouldn’t be surprised if it starts sooner. That means companies need to be looking both at near term details and long term changes to the industry.
When I asked about long term strategy, I got the typical startup answer: They’re focused on internal growth rather than acquisition (either direction). That’s a good external message because folks who want a leading edge company want it clear that they’re using a leading edge company, but I hope the internal conversations at the CxO level aren’t avoiding acquisition. That’s not a failure, just a different version of success.
Denodo is a strong technical company focused on data virtualization in the short run. They have a very nicely scaled model from Denodo Express to their full product. They seem to understand their sweet spot within IT organizations. Given that, any large organization looking to get better access to disparate sources of data should talk with Denodo as part of their evaluation process.
My only questions are in marketing messages and whether or not Denodo be able to change from a technical sales to a higher level, clearer vision that will help them cross the chasm. If not, I don’t think their product is going anywhere, someone will acquire them. Regardless, Denodo seems to be a strong choice to look at to address data access and integration issues.
Data virtualization is an important niche, the questions remain as to how large is the niche and how long it will remain independent.
I watched a webinar yesterday. It was sponsored by Magnitude, the company that is the result of combining Kalido and Noetix. The speakers were Ralph Hughes, a data warehousing consultant operating as Ceregenics, and John Evans of Magnitude.
Ralph Hughes’ portion of the presentation was very interesting in a great way. Rather than talking about the generalities of enterprise data warehouses (EDW) and Agile, he was the rare presenter who discussed things clearly and in enough detail for coherent thought. It was refreshing to hear more than the usually tap dance.
Ralph’s slide on the advantages of agile development for EDW’s is simple and clear. The point is that you don’t know everything when you first acquire requirements and then design a system. In the waterfall approach, much of coding, testing and usage is wasted time as you find out you need to do extra work for new requirements that pop up. Agile allows business users to quickly see concepts and rethink their approaches, saving both time to some productivity and overall time and effort of development.
After talking about agile for a bit, he pointed out that it does save some time but still leaves lots of basic work to do. He then shifted to discuss Kalido as a way to automate some of the EDW development tasks in order to save even more time. He used more of his presentation to describe how he’s used the tool at clients to speed up creation of data warehouses.
One thing he did better in voice than on slides was to point out that automation in EDW doesn’t mean replacing IT staff. Rather, appropriately used, it allows developers to move past the repetitive tasks and focus on working with the business users to ensure that key data is encapsulated into the EDW so business intelligence can be done. Another key area he said automation can’t do well is to manage derived tables. That still requires developers to extract information, create the processes for creating the tables, then moving the tables back into the EDW to, again, enhance BI.
Notice that while Mr. Hughes spoke to the specifics of creating EDWs, he always presented them in context of getting information out. Many technical folks spend too much time focused on what it takes to build the EDW, not why it’s being build. His reminders were again key.
John Evans’ presentation was brief, as I always like to see from the vendors, rounding out what his guest speaker said. He had three main points.
First, the three main issues facing IT in the market are: Time to value, time to respond to change and total cost of ownership. No surprise, he discussed how Magnitude can address those.
Second, within his architecture slide, he focused on metadata and what he said was strong support for master data and metadata management. Given the brief time allotted, it was allusion to the strengths, but the fact that he spoke to it was a good statement of the company’s interests.
Third, he discussed the typical customer stories and how much time the products saved.
The webinar was very good exposure to concepts for an audience thinking about how to move forward in data warehousing, whether to build EDWs or maintain them. How agile development and an automation tool can help IT better focus on business issues and more quickly provide business benefit was a story told well.
Last Friday’s BBBT presentation was by Michael Whitehead, CEO, WhereScape. The company seems to have a very interesting and useful product, but there’s a huge communications gap that needs to be addressed.
What They Do
One marketing issue to start was that I got most of this section from my own experience and WhereScape’s web site, not from Michael’s presentation. When someone begins a presentation by proudly announcing it is ““guaranteed there’s no corporate marketing in the presentation at all” while you’re presenting to a group of analysts, there’s a disconnect and it shows.
WhereScape has two products, Red and 3D, to help build and maintain data structures. The message is focused on data warehouses, but I’ll discuss that more in the next section. One issue was that their demonstration didn’t work as there seemed to be a problem connecting between their tablet and the BBBT display system, so much of what I’m saying is theory rather than anything demonstrated.
Red is their tool to build data warehouses. Other tools exist and have been around for decades, Informatica being just one competing firm.
3D is where the differentiation comes in. Everyone in IT understands that nightmare that is upgrading major software installations such as ERP, CRM and EDW systems. Even migrating from one version to the next of a single vendor can involve months of planning, testing and building, followed by more months of parallel runs to be safe. A better way of analyzing and modifying data structures that can compress the time frame can have a large positive impact upon a corporation. That’s what WhereScape is attempting.
What They Say
However, their message is all “Automation! Automation! Automation!” and the short part of the demo that worked showed some automated analysis but a lot of clicks necessary to accomplish the task. From what I saw, it will definitely speed up the tasks, if as advertised, with clear time and money savings, but it’s not as automated as implied and I think a better message is needed.
In addition, their message is focused on data warehouses while Michael said “We’re in the automation business not the data warehouse business,” which really doesn’t say anything.
Michael did talk for a bit about the bigger data picture that includes data warehouses as part of the full solution, but again there’s no clear message. While saying that he doesn’t like the term Data Lake, he’s another that can’t admit that it’s just the ODS. There’s also a discussion of the logical data warehouse, also not something new.
One critical and important thing Mr. Whitehead mentioned was something I’ve heard from a few people recently, the point that Hadoop and other “unstructured databases” aren’t really unstructured, they support late binding, the ability to not have to define a structure a priori but to get the data and then understand and define a useable structure for analysis.
What They Need to Say
This is the tough one and not something I’m going to solve in a short column. The company is targeting a sweet spot. Data access has exploded and that includes EDW’s not going away, the misnamed concept of Big Data and much more. Many products have been created to build databases to manage that data but the business intelligence industry is still in the place packaged, back-end systems were in the 1990s. Building is easier than maintaining and upgrading. A firm that can help IT manage those tasks in an efficient, affordable and accurate way will do well.
WhereScape seems to be aimed at that. However, their existing two-fold focus on automation and data warehousing is wrong. First, it doesn’t seem all that automated yet and, even if it was, automation is the tool rather than the benefit. They need to focus on the ROI that the automation presents IT. Second, from what was discussed the application has wider applicability than just EDW’s. It can address data management issues for a wider area of business intelligence sources and the message needs to include that.
Though the presentation was very disjointed, WhereScape seems to have focused on a clearly relevant and necessary niche in the market: How to better maintain and upgrade the major data sources needed to gain business understanding.
Right now, while there is a marketing staff at the company, WhereScape’s message seems to be solely coming from the co-founder and CEO. While that was ok in the very early days, they have some good customer stories, having led with Tesco’s success in this presentation, and it’s time to leverage a stronger and clearer core message to the market.
Where the issue seems to be is the problem I’ve repeatedly seen about messaging. The speed of the industry has increased and business intelligence is, on a whole, crossing Jeffrey Moore’s chasm. That means even younger firms need to transition from a startup, technically focused, message to a broader one much more rapidly than vendors needed to do so in the past.
While WhereScape has what seems to be the strong underpinnings of a successful product, they need to do some seriously brainstorming in order to clarify and incorporate a business oriented messaged throughout their communications channels – including in presentations by founders.
Data warehouse theory originally posited extracting data from systems, performing transformations on them and loading the resulting schemas into the data warehouse. It was a straight flow of information. However, the difference between theory and practice quickly reared its head. Today, people are talking about Data Lakes and Data Swamps. They’re not new, they’re just the ODS updated for modern data.
Data Warehouses and the ODS
Academics don’t have to deal with operational systems. In the 1980s and 1990s, those systems were growing, with ERP, CRM and other systems increasing the complexity and volume of data. Those mission critical systems, however, weren’t designed for extraction of information. They were primarily running on RDBMS systems that had locking schemas that could grind process and transactional systems to a halt while and extraction program kept open large blocks of records while transforming basic data in star schemas. Something needed to be done.
There was also a secondary effect that was very important to some people. IT, just as with ever other department in a large enterprise, isn’t monolithic. The people managing the operational systems knew their systems were mission critical and also knew how, in reality, those systems were big but fragile. They weren’t happy with opening their operational systems to other IT folks who were interested in non-operational things. Those folks answering other business problems? They were viewed as intruders, getting in the way of the “real work.”
For both reasons, intrusions into the operational systems were something to be kept to a minimum. IT organizations began using an Operational Data Store (ODS) to quickly open the operational systems, suck all the data out, willy-nilly (yes, I decided to use that term in a tech article…), and then go back to prime performance in an isolated system.
It was then the ODS that was the source of the data warehouse ETL process. On a tangent, this is why the people now arguing about ETL v ELT amuse me. It’s been ELETL for decade, if we want to be honest; but who cares? I’d rather have a BLT than spend so many cycles over slightly different acronyms for concepts that ETL handily describes, even in permutations.
The ODS comes into its own
The IT folks who were working to provide reports for mid- and high-level managers were always trying to tweak enterprise software reports, trying to extract nuggets of value. The data warehouse was a step forward and helped build a bigger picture. However, the creation of star schemas and other DW techniques aggregated data and lots a lot of detail. A manager would see an issue and want to backtrack, to drill-down into the data to know more.
The ODS became the way to do so. Very quickly, the focus changed from ODS in front of the data warehouse to both working side-by-side. Having all that raw data available gave the business analysts a way of providing much more detail and information to the business user. The first big BI companies, those such as Cognos, Business Objects and more, leveraged the two data stores to provide an ability to drill down past the aggregate information into the more detailed data.
Having that large volume of data from multiple operational systems also intrigued people who weren’t data warehouse focused. They wanted to sift the raw data for technical or performance trends, things that weren’t of interest to the typical DW designers and users, but were important to mid-level management in manufacturing, marketing and other departments. Business analysts supporting those people began to turn to more and more analysis directly on the ODS data
The ODS comes to the fore – by another name
That was happening in the 1990s, at the same time another key phenomenon was growing: The Web. The growth of the web meant a lot more data about a lot more things. Web sites are operational systems to marketing in just as critical a way as an assembly line is to manufacturing. People became interested in ensuring that what visitors to web sites did was captured and available for analysis. However, as the volume of web traffic grew exponentially, new issues had to be looked at to handle that data.
Columnar databases were one solution, a way to speed up analysis of dimensions of information across individual records. The vastly larger amount of data also helped push emerging MPP technologies and drove creation of Hadoop and other technologies that could manage much larger data sources much faster and more cost efficiently than could individual Unix servers.
However, the web folks were new to IT and grew up in a different generation than the folks who designed and drove data warehousing. It’s natural to want to take ownership of concepts, especially those on the edge. So the folks working with these new data sources began talking about Big Data as somehow completely different than what came before. If that was the case, they needed to think of some term for the database where they dumped all the data extracted from web sites. Data Lakes became one term. We’ve heard data swamp and other attempts to create unique terms so a company can differentiate itself from others. However, there’s already a name.
The ODS exists. It’s evolved. It’s moved forward. But it’s still the ODS.
“But,” you say, “ an ODS is operational information and the data lake is so much more!” Well, not quite. There are two main problems with that argument.
First, times change. When the ODS was coined, the focus was on the back-end systems such as ERP, CRM, accounting and other fairly closed systems. It was before the web, before the ubiquity of mobile devices, before the wall between back-end and customer-facing systems was destroyed.
As mentioned, not just web sites are but even the internet is an operational system for your business – and not just for ecommerce companies. From lead generation, to maintenance and training, the internet is a key tool for providing operational support and generating business critical operational details.
Second, just as ETL can mean a number of things, so can ODS extend past a pure theory while still being relevant. CRM systems are considered operational but still contained sentiment and other information in comments fields. Just so, the vast volume of data from a call center’s voice recording system being dumped into the ODS have two components. There are basic details about the operation of the call center, things such as number of calls, call length and other details that are purely operational. There are also additional details about customers that can be distilled for strategy purposes, including the ability to provide sentiment analysis. Just because an operational system captures data that can be used for more than purely operational decision making doesn’t obviate that the information extracted resided in an ODS.
It’s a need of information technologists from all generations to realize that things change but retain context. The ODS isn’t what it was thirty years ago, but the data lake also isn’t some new creation born full blown from the web. There are few truly revolutionary technologies. You can be a brilliant person and contribute much to technology and business and still not be a revolutionary.
The ability to manage the vastly larger amounts of data than we had twenty years ago is critical. There are many innovative things being done. However, I consider the first expert systems, the first MPP algorithms and other similar technologies to be revolutionary. The fact that what is being done to allow business to gain insight combining more and more data from even more diverse sources is no less valuable to the industry because it is instead an evolutionary change.
The ODS has evolved. It doesn’t need a new name, just a tad more respect.
The latest presentation at the BBBT was by Amit Bendov, CEO, Sisense. First marketing warning: If you’re going to their web site, be prepared. Maybe it’s only for some weird Halloween thing, but the yellow and black background of the web site is the one of the ugliest thing I’ve seen for a professional company. However, let’s look under the covers, because it gets better.
The company was founded in 2004 and Amit says the first sales were in 2010. There’s a good reason for that delay. They are yet another young company who talks about being a full stack BI provider, being more than a visualization tool but also supposedly providing ETL, data storage and the full flow for your information supply chain from source systems to display. That technology took a while to develop.
Technology: Better integration of memory and Disk
The heart of their system is a patent pending technology that tightly integrates cpu cache, RAM and disk to better leverage all storage methods for higher performance. The opportunities that theory provides are enough that they’ve received $50 million (USD) in venture funding, $30 million in their latest round, earlier this year.
As they are a startup, it’s no surprise that the case studies given were for SMB or departments within enterprises. That’s the normal pattern, where a smaller group takes advantage of flexibility to try new products to solve focused problems. As their customer list includes companies such as Ebay, Wix, ESPN and Merck, companies with lots of data, those early entrants increase the potential if Sisense continues to perform.
Another key technology component is their columnar database. They created a proprietary one to be able to support their management technology. That’s completely understandable as their database isn’t purely on disk or memory, but in a combined mix that needs special database management.
The final key to their technology is that they worked to ensure the software runs on commodity chips from the X86 heritage. That means it runs on normal, affordable, off the shelf servers, not on high priced appliances.
The combination of the speed and affordability of the technology is justification for the rounds of funding they’ve received.
Really full stack?
One fuzziness that I’ve mentioned with other full stack vendors is the ETL side of the process. The growth of Cloud companies such as Salesforce, and the accessibility of their APIs, means that you can get a lot of information out of systems aimed at SMB. However, true enterprise ETL means accessing a very wide variety of systems with much less easy or open APIs. When Mr. Bendov talked about multiple systems, it seems, from presentation and demo, that he’s talking about multiple instances of simple databases or open APIs, and not a breadth of source types. There wasn’t a lot of choice in the connection section of his application.
That’s not a problem for companies at Sisense’s state of maturity, as long as there’s a business plan to expand to more enterprise sources. They need to focus on proving the technology in the short term and having more heterogeneous access in their tool bag for the future.
Another issue is the question of what, exactly, their database is. Amit Bendov made a brief comment about not needed data warehouse, but as I and others quickly brought up, there are two problems with that statement. First, they would seem to be a data warehouse. They’re extracting information from source systems, transforming that information even if not into the old star-schema structures, and providing the aggregate information for analysis. Isn’t that a high level description of a warehouse? Second, as they’re young and focused on SMB or departments, as with other companies who serve visualization, they might need to look at customer demands and get access to corporate data warehouses as another source.
The old definition of a federated data warehouse seems to be evolving into today’s environment where sometimes an EDW is a source, other times a result and sometimes it’s made up of multiple accessible components such as Sisense and other databases. Younger companies who disparage EDWs need to be careful if they wish to address the enterprise market. The EDW is evolving, not dying off.
User interface and more
One of my first trips to Israel was, in part, when my boss and I had to bring a couple of UI specialists to show Mercury Interactive’s programmers why it might be nice to rethink application interfaces. It’s wonderful what twenty years have wrought. Amit Bendov says that Sisense has one UI specialist for every two programmers, and the user interface shows that. While I mentioned that they need broader ETL access, the simplicity of getting to sources is clear. While you still will need a business analyst to understand some column names, it’s a very easy to use interface.
The same is true in the visualization portions of their application. While it’s still a simpler tool, it has all the basics and is very clear to understand and use.
Paving the way for their spread into enterprise, the Sisense team also supports single-sign on, basic data access control, both in global administration and in the user interface, and other things that will be needed to convince a larger corporation to spread the technology.
Sisense looks like a startup in a great position. Their technology is well thought out and seems to be performing very well in the early stages. Affordable, fast, business intelligence is something nobody will turn down.
The challenge is two-fold:
- Do they have the technology plans to help them address larger enterprise issues?
- Do they have the mindset to understand the importance not only in marketing, but in changing the marketing to a more business focus?
This is the same refrain you’ve heard from me before and which you’ll hear again. This is the Chasm challenge. Their technology has a great start, but their web site and presentation show they aren’t yet thinking bigger and we’ll have to see what the future holds both for the technology and the messaging.
Business intelligence is a very visible market and one growing quickly. While small companies need to focus on the early adopters, they must very rapidly learn how to address the enterprise, both in products and marketing.
High performance BI at a reasonable cost is a great sell, but Sisense isn’t yet read for full enterprise. Sisense has a great start but life is fluid.