Africa and the next generation of database systems

Africa and the next generation of database systems

IT veteran and entrepreneur Barry Morris, co-founder and CEO of NuoDB, an established operator in scale-out SQL database technology, offers his insight into re-defined relational database technology and its role in tackling multiple data challenges faced by businesses.

Morris speaks to ITWeb Africa about the implications of businesses across the continent being motivated to move towards commodity data centres, and the real opportunities that lie before those who operate within data-centric business environments - who both understand and embrace the evolution of database management systems.

Chris Tredger: What is relational database technology?

Barry Morris: Relational Databases are used for managing valuable data. It's how we run the world, and how we have done so for 30 years.

In the 21st century everything is digital, and digital is data. There is a reason we call our industry Information Technology: It is about managing that data. And the most important, valuable data is managed using RDBMS (Relational Database Management System) technology.

The largest technology companies in the world are RDBMS vendors, including ORACLE, Microsoft and IBM. Their RDBMS products manage your checking account, your company's inventory, the booking systems for your airline flights, your credit card transactions, your tax records, and every other digital thing that involves valuable data.

The underlying technology is not particularly interesting to the man in the street, but the capabilities are critically important. An RDBMS is a highly reliable system that guarantees things about the data, for example that when you deposit money in your bank account it is still there tomorrow, or that if two people try to withdraw the last cash from an account at the same time at most one person will get that cash.

There is a lot of detail about the technology, including that the data is stored in tables or Relations (hence "Relational Database"), and there is a whole mathematical model behind this based on formal set theory, but to the non-expert the key point is that some data has to be stored reliably, manipulated very flexibly and accessed very rapidly. When those are the requirements we turn to an RDBMS.

Chris Tredger: Why should businesses in Africa care?

Barry Morris: Businesses in Africa care about Relational Databases because they are businesses. They have valuable data, ranging from customer data to product data, to transaction data, to employee data, and more. This is all stored in Relational Databases today, and will continue to be for the foreseeable future. Of course they may have lots of data that is valuable but not business critical, like the temperature history of the storage warehouse, or the number of visitors to their website. That data may or may not be stored in a Relational Database system.

The challenge with Relational Database technology is that it is essentially 1980s technology. It is a surprise to modern IT folks to hear that these systems were originally built for databases smaller than 1GB, running on 32-bit machines, with transaction rates much lower than 100 TPS, and a few hundred concurrent users.

Most critically they were designed to run a single database on a single dedicated server. The systems have evolved a great deal over the years, but they are still really based on the same single-server storage-centric architectures. The world has moved on to run on gangs of inexpensive commodity servers with huge DRAM configurations, to private clouds, public clouds, web-scale concurrency requirements and, of course, to very large database sizes.

After 30 years the old architectures for Relational Databases are creaking and expensive.

In the African context businesses are especially motivated to move to commodity data-centres, primarily for cost reasons. These servers cost much less than R10,000 each and each server can be virtualized to run multiple workloads simultaneously. It is enormously less costly to run our IT workloads on a uniform set of these machines.

The machines cost less, the management is easier, each machine is utilized more effectively, and the whole data-centre is greatly simplified. Critically the uniform nature of these environments also reduces the dependency on specialized expertise across the wide range of technologies that are deployed in conventional data-centres.

Whether a particular business thinks of this as a private cloud, a scale-out data-centre, or a virtualized environment it boils down to the same thing: Huge efficiencies. And for those that are prepared to rent some of that capability on so-called public clouds, like Amazon EC2 or IBM Softlayer, there are even bigger savings to be made.

Africa can leverage all of this because it allows large companies to scale-up without the IT cost complexities, and it allows small companies to get going without the huge capital outlay historically required to run a conventional data-centre infrastructure. But there is a problem, namely that conventional RDBMS technology is poorly suited to these new data-centre environments. It can be jammed in, at great expense, but a new architecture for the RDBMS that is designed "On the Cloud, For the Cloud" would the big enabler for next generation IT.

Chris Tredger: What are the main challenges impacting on database technology application and experience in the workplace?

Barry Morris: We live in exciting times: Web, mobile, social networks, and devices in our pockets that give us 24x7 access to everything: photos, SMS, location, email, entertainment media etc. It's hard to look in any direction without seeing new mobile payment technologies, keep fit technologies, ways to buy pet food or ways to enjoy our friends' vacation experiences.

A parallel revolution is happening in businesses, where everything is becoming data-centric and information driven. In short we have moved into the world of Being Digital, as predicted in 1995 by Nicholas Negroponte of the MIT Media Lab.

What it means to IT people is that we're facing a data deluge. The Volume, variety and velocity of data is off the charts compared to where it was when Negroponte wrote the book. But so is the Value of the data.

Companies that don't understand the Value of data will be put out of business by aggressive data-driven competitors. That means every business in under pressure to put in place the strategies, technologies, processes and people to be data-centric.

Our focus is the technology side, of course. When someone tells me that they need to cope with millions of concurrent web users, or tells me that they have run their systems in an always-on, 24x365 fashion, or tells me that they need to give their users highly responsive access to information in Johannesburg, Cape Town and Durban simultaneously, I conclude that they are building a modern business. But I also know that the limitation will be about how to manage the data efficiently and cost-effectively. It is a job for a next generation database system.

Of course there are also skills issues. One of the key ideas in the world of Relational Database systems is to have the database do all the work. You don't have to tell these systems how to do things, you just tell them what you want them to do.

Folks familiar with the SQL language will know what I am talking about. You tell us what data you want and we'll get it to you in the most efficient way. You tell us to update a lot of complex data and we'll do it reliably or tell you why it can't be done. That's how it works. And it's what you want so that everyone from your programmers to your administrators can get on with their jobs and leave us to do the complicated detail stuff about guaranteeing he safety and integrity of your data.

The best news is that there is no re-training required. IT people already know how to use these systems. What has changed is how we do it behind the scenes, not how you tell us what to do. It's like a car: You don't care what kind of engine it has, as long as the steering wheel, pedals and other controls work in the conventional fashion. We don't need retraining when we drive a new car (although personally I would prefer them to pick a side for the direction flashers and stick with it).

The bottom line is that we need to reduce the data management skills requirements over time so that businesses can focus on what they do well, namely selling things, constructing things, providing services etc.

Chris Tredger: Is Africa keeping up with growth of database management trends, technology? what more needs to be done to improve the continent's overall capability?

Barry Morris: Africa needs to be a fast-follower when it comes to database technologies, and other technologies too. There are plenty of really smart people in Africa that could strive to be pioneers, but in a resource and expertise constrained environment it is better to watch the pioneers closely and then commit to successful directions when they emerge. Fast followers can be very efficient, and that is the right way for Africa and African organizations to behave in most cases.

Is Africa keeping up with Data Management trends? My answer is yes – absolutely. When I look at the number of folks that we are talking to across Africa, and specifically in South Africa it is remarkably high relative to other regions of the world. We have just set up a South African partner to represent us in the region, and I was thrilled to be able to show them the level of interest in NuoDB.

It is my view that Africa is very switched on to opportunities for technological advantage, and I believe the continent is well placed to take advantage of the commodity data-centre revolution, along with the data management implications. Naturally that is why NuoDB is investing at this point.

Chris Tredger: Is there an appreciation for what databases can do, how they can be leveraged to enhance operations?

Barry Morris: Absolutely. Having said that, it is hard not to understand that once you embark on a data-centric corporate strategy. If you want to store grain you need silos, and if you want to store data you need databases. The latter requirements are more nuanced or course, including wide variety of things that you want to do with stored data and the importance of not losing a single "grain" of it.

But the principle stands – people understand that they need to put data at the centre of how they run their businesses, and they quickly grasp that a next generation data management system is a key part of doing that.

Chris Tredger: Security remains a critical issue in Africa, particularly with regard to data-intensive processes and management issues - what is your view?

Barry Morris: It's true in Africa, but no more so than anywhere else in the world.

Again, let's distinguish valuable data from data in general. With some data you really don't care too much about security. Usually auto-generated data is least valuable, and usually personal and transactional data is the most valuable. But in general no-one wants any data compromised, and security is a challenge across the board.

The obvious point is that technology is an enabler but it's never the whole solution. I can put as many locks on my house as a like but if I don't look after the keys then the locks don't guarantee very much. The same is true of digital security strategies. The technologies need to be there, but they need to comprehensive security processes that are designed well, governed by rigorous processes, and managed with discipline.

On the technology side of security in data management it gets pretty technical pretty quickly. I can tell you that we require all machines to be mutually authenticated, that nodes participating in a database have to have the right credentials to do so, that all communications are encrypted, that we support multi-key encryption strategies for data at rest, and so on. That makes NuoDB very secure, and you can rest easily that your data is safe. But if you leave the front door wide open then people can walk into your house and take what they want. You can't blame the guy that built the house for that!

In general I think that the move to commodity data-centres is going to simplify the data security problem, because it simplifies the whole data-centre. Organisations will have fewer security holes to worry about and will do a better job of protecting themselves.

The architects of fortified cities and castles made sure that there were a small number of entry portals and they protected that small number of "weaknesses" with drawbridges, portcullises and related approaches. It's the same for data security: Simplify the problem down to a manageable number of vulnerabilities and then put your energy into securing those. For sure commodity computing and scale-out data-centres will do that for us.

None of this is Africa-specific, but Africa certainly benefits from the simplification because there are relatively fewer security experts in Africa than Western Europe of the USA, and African organisations therefore have a disadvantage in terms of securing more complex data-centre infrastructures.

Read more