Ingres recently announced a name change to Actian. This raised a number of questions which Nikos Vaggalis put to Roy Hann, chair of the UK Ingres Users Association.
Why the name change from IngresCorp to Actian?
I have expected a name change for quite some time now. The company has a number of very strong products in addition to the Ingres DBMS. For instance there is its spectacular high-performance Vectorwise analytics engine that it launched just over a year ago. It doesn't make sense to shine the spotlight on just one product in a growing range of complementary products.
I assume you are asking about the new Action Apps and the Cloud Action Platform. I'd say it indicates expanded ambition and growing confidence in the ability of Vectorwise to power novel solutions as well as traditional business intelligence tools. One thing that has become clear to me using Vectorwise over the last two years is that conventional BI tools get lots of benefit from Vectorwise but the tools hold Vectorwise back. It would shine a lot brighter if it wasn't hidden under layers of stale old work-arounds for the limitations of older, frankly unsuitable database engines. Action Apps are one example of new kind of tool that is made feasible by "near time" querying of colossal databases using Vectorwise.
Does it indicate a shift in priorities and a new focus on a particular direction ?
What changes do you anticipate in product development?
The Action Apps and the Cloud Action Platform is going to be a whole additional area of development. Looking at development efforts on the other products, the Vectorwise team is working flat-out adding new features. In particular they are working to add support for some of the more advanced analytic capabilities required by recent SQL standards, such as windowing functions. If we look at Ingres Database 10, it was released about a year and half ago and a lot of effort is going into stabilising the major new features it introduced, such as
its very slick new MVCC implementation. Its capabilities are now well ahead of where the majority of customers are. Also, while all that work is going on, Ingres Geospatial is nearing completion and is expected to be released later this year.
All of these products benefit each other. It's a virtuous cycle. Action Apps will depend on Vectorwise; Vectorwise needs Ingres; Ingres is very easy to introduce into a site that has adopted the unique capabilities of Vectorwise and Action Apps. It all makes sense.
Will Ingres and VectorWise join code bases and what will this mean for existing products?
Yes, the project to merge the code bases has been underway for several months already. The engineering teams certainly intend to keep the code lines coherent as far as possible, though of course there are features of Vectorwise that make little sense to add to Ingres, and
Having said that, there are enhancements required by
Vectorwise that Ingres will benefit from. For instance the bulk load utility has been recently re-worked and both products get the benefit. Similarly the management tools are being overhauled and made easier to integrate and again both products will benefit. What's good for Vectorwise is usually good for Ingres too.
Well first of all, apart from Vectorwise and Ingres, my company, Rational Commerce, is entirely product neutral. Different customers have different needs and
How are Ingres-VectorWise related to other components of a BI solution?
preferences when it comes to BI tools, but thanks to Ingres' decades-long commitment to open standards almost any BI tool you could name is fully supported and certified with Vectorwise and Ingres. Incidentally, it's not just the usual business-oriented tools that work well with Vectorwise; for example it also works brilliantly with the R programming language which is emerging as the de facto standard among scientists, engineers and statisticians. In fact all the big name stats packages can exploit Vectorwise.
Great question; it really couldn't be simpler! Vectorwise gives results that are within a whisker of the published TPC-H benchmarks on relatively low-end, commodity hardware. I have a machine on my desk right now that cost less than £6,000 and delivers more than 90% of
What are the requirements (for example in hardware or infrastructure) needed for deploying your BI solution?
the published TPC-H performance at any given scaling factor. Just astonishing.
Having said that, we've got a system on which we did some fine tuning and spent a lot of time selecting suitable matched components. It ended up costing exactly the same but it delivers more than 95% of the
published TCP-H benchmark performance. We're selling them as turnkey bundles with Vectorwise installed, with Windows or Linux. Plug them in and start using them.
It's almost as easy if you already have your ownhardware. You need a 64-bit Intel or AMD processor, with as many cores as you like, and at least 32Gb of RAM, and that's about it. Vectorwise does not benefit
from fancy disk subsystems. The software takes about 10 minutes to install. It uses absolutely standard SQL so you can be productive immediately.
There is also a multi-tenanted Vectorwise cloud offering from Actian.I must admit I don't know too much about hosting databases in the public cloud, but I can certainly see how it'd be a great way to aggregate small volumes of data for vast numbers of users, like say smart utility meters.
We have been doing some work recently that uses the cloud the other way round. The BI tools live in the cloud but the database stays inside the customer's firewall. Users are able to access the BI tools and reports on demand, and queries are sent to the customer's database from the cloud. Results are presented back to the users. It's a great way to put tools in the hands of users as and when (and where) required without shifting vast mountains of sensitive data over the Internet.
With Vectorwise that can be pretty simple. I mentioned
After deployment, how do I get started with collecting, mining and finally interpreting and taking advantage of that data ?
above that it doesn't need work-arounds like star-schemas and dimensional models. Those are used to get adequate performance out of conventional row-oriented database engines. That whole layer of
analysis and design and data transformation contributes no value; it is there just to make the thing work acceptably quickly, and it costs a lot of money to set up and to run. With Vectorwise you can often take your production transactional database design and simply copy the required data, as-is.
This has several obvious advantages. It's simple. The database can be refreshed quickly and therefore frequently. The data model is easily comprehensible to the less technical, more business-aware user. It is
And that's the key. To take advantage of the data you need to know what the business needs to know. Breaking down the barriers between the business data and the business people means they make better decisions faster.
I expect Action Apps will be used alongside any existing BI solutions quite happily. They address different needs. Conventional BI applications tend to be deeply retrospective whereas Action Apps are expected to repeatedly poll streams of incoming, often raw data at short intervals. I imagine customers will find Action Apps invaluable for running their business day-to-day,
How will switching to Actian Apps affect existing infrastructure?
moment-to-moment while any existing conventional BI tools will be used much as they use them today.
Why should a business prefer your BI solution over a competitor's?
It's cheap and it's fast. Very very fast. Also the minimum price of entry is very low. You don't need to scale up to a huge server farm to get low cost per query. You can query billions of rows in under a minute on even a single, ordinary office PC. You can do more queries in the the time it used to take to run one query. You can try many
Is the push for better and faster BI tools because of a shortage in data analysts?
different ideas in the same time it took to try one . You can explore. You can have more eyes looking at the data in the same time. And you can work with fresh, up-to-the-minute data because you don't need to go through a tedious, slow transformation process. You can use the
raw data as quickly as you can load it. Finally you can implement Action Apps so that you can be alerted to changing business conditions in near real time. It's weather radar for the stormy business environment.
There really is no down side.
That, and what they do has not delivered the ROI that was perhaps expected. It's not their fault. The existing tools are constrained by the state of the database art fifteen or twenty years ago. SQL databases took a serious wrong turn almost from Day 1 by implementing rows as if they were records in a file. Performance was bound to be poor.
In time the data analyst's role will change. For now I think the whole approach to BI is shaped and constrained by the limitations of some very old technology trends that are going to take a while to shake off. In the longer term I think the data analysts will be able to spend more time on the more important questions they deal with, like data collection protocols, data quality, harmonising representations, making clearer distinctions between incomparable data, and most importantly, on figuring out what the business needs to know and delivering it promptly. In future I hope they will spend a lot less time
Do these tools replace the data analyst or do they complement his work ?
on endlessly re-arranging the data to optimize query performance. That is just money down the drain.
I hope that by using a more natural model for the data it will become directly accessible to users who actually run the business. They will still need to understand the technology quite well, but it will require only genuinely useful technical literacy, rather than silly stuff like knowing how to bodge the data into variouswork-arounds.
Can anyone use these tools or is a skilled professional still required?
Yes. But to be more specific, they'll need strong knowledge of the analytical capabilities of SQL. They won't need SQL to do updates, nor transaction design, nor even table design. It is non-trivial but could be mastered in a week or two.
So what skills should a modern data analyst have in his armory? Strong SQL and Relational/Database skills?
Finally can you give us a sneak preview of the next UK Ingres Users Association gathering ? Is it going to revolve around Actian Apps?
First of all I want to emphasise that the UK IUA is entirely
self-funded and independent of Actian. The UK IUA is a friend of Actian but it exists to serve and promote the ordinary members' interests. It is not a marketing tool for the company and it is a very broad church.
The detail of the 2012 UK IUA conference won't be decided until about March, but we do allow the event's major sponsor to set the theme of the morning plenary sessions. If Actian is the major sponsor in 2012, I
imagine they will give some prominence to Action Apps and the Cloud Action Platform, along with their other products. We have not yet found a major sponsor for 2012.
The content of the afternoon technical break out sessions is chosen by customers and partners, so I would expect the usual very strong emphasis on OpenROAD, along with Ingres and Vectorwise. If there are people who want to present what they've been doing with Action Apps that will be great, but it will be one track among many.
Roy Hann is Technical Director of Rational Commerce Limited, an Actian Partner specialising in Vectorwise and Ingres solutions. He has been the chair of the UK Ingres Users Association for three years and has also been a member of the board of the North American Ingres Users Association. Roy blogs about Vectorwise and speaks at Ingres conferences in Europe and North America. He has also designed a series of amusing T-shirts for the UK Ingres development sprints, sales of which help fund the UK IUA.