Teradata switches CEOs mid-flight while being eaten alive in the cloud, but it's not game over yet for data warehouser

Revenue's tumbling, but that's also typical of a move to the fluffy stuff

Analysis When Oliver Ratzesberger was named as Teradata CEO and president in January last year, it seemed he was made for the job. Not only did he have a background in engineering and software, but thanks to seven years leading analytics at eBay, he was also well versed in the challenges of big data. By December, he was gone.

Last week, the data warehouse company announced Steve McMillan as president and CEO. He will start work in June following Victor Lund's caretaker period in charge.

McMillan will have nerves to steady and stories to set straight. He must placate employees, investors, and customers. After all, there is a lot to explain if you've got through three CEOs in 18 months.

First of all, there are the numbers. Ratzesberger's departure was prompted by disappointing results in calendar Q3 2019, when revenue dropped 11 per cent year-on-year to $459m. In figures released last week (PDF), Teradata's Q1 of 2020 showed revenue was down again, this time by 6 per cent to $434m, including some negative impact from COVID-19. (Net income was up to $168m after reporting a loss of $10m in the same quarter last year.) The trouble for Teradata is that falling revenue has become a long-term trend.

Big data and the move to the cloud

The company's struggles are the result of a two-pronged assault on its market. Firstly, Terradata's relational data tech has seemed a bit old hat during the growth in popularity of big data technologies – that is the fast-moving, unstructured data spewing from online content, smartphones, and IoT. And Hadoop, the distributed file system once the cornerstone of any self-respecting big data stack, has itself been superseded by object storage from the cloud vendors - such as AWS's S3 - as the growth in the family of NoSQL databases continues unabated.

Then there is the cloud itself. Gobbling up a huge volume of the hosting and infrastructure market, the big three cloud vendors have wasted no time in getting stuck into the data warehouse market too. AWS has Redshift, Azure has Synapse, and Google has BigQuery. Then there are the cloud-native vendors such as Snowflake.

Last year, Gartner noted that just two vendors – AWS with Redshift and Microsoft with Azure SQL Data Warehouse – accounted for 70 per cent of the revenue growth in the overall DBMS market for 2017 and 2018.

Carl Olofson, research veep at IDC, said that Teradata's revenue was bound to take a hit as it shifted to monthly or quarterly cloud services contracts from a model that was based on upfront purchases of appliance servers and specialist software.

"When you move from the old appliance on-prem data warehouse model to the cloud model, of course, hardware income drops off significantly, and you also see the charging models shift from a perpetual-use software licence model to a cloud subscription model. And when that happens, there's always a dip in revenue and every company that's gone through this transition has seen that: it's unavoidable. But it can work out better for the software organisation."

The view is borne out in Teradata's Q1 2020 results [PDF], which showed a staggering 55 per cent annual drop in perpetual licences and hardware. Recurring revenue, which includes subscription licences, grew 4 per cent and made up nearly 80 per cent of total quarterly revenue.

But the results for consulting, which fell 29 per cent, are perhaps more of a worry, and why the company picked the new CEO.

Before agreeing to take the helm at Teradata, McMillan served as executive vice president of global services at F5 Networks. He led its worldwide services operations, including global support and consulting. He also had spells at Oracle and IBM, two Teradata competitors in the enterprise data warehouse market, where he held services roles.

Olofson said: "The new man has got a very strong background in service delivery. And that's certainly an important part of the story because enterprise data warehouse is a high-touch environment and people need to know what they're doing to be able to do this right. They are necessarily going to have to have a strong services component. [The new CEO] sends a message that they're intending to ensure that their customers have the services in terms of consultants, guides and training."

Even so, McMillan will have to deal with the fact that the cloud-native competition has already started eating Teradata's lunch.

Rivals

Consumer goods giant Unilever, which began a shift to a global enterprise data warehouse on Teradata back in 2010, started flirting with Azure's Synapse during 2019.

"We are very excited that Azure Synapse Analytics will streamline our analytics processes even further with the seamless integration the way all the pieces have come together so well," Nallan Sriraman, global head of technology at Unilever, said at the time in a canned statement.

Teradata assured The Register that Unilever is still a Teradata customer and uses its Vantage platform for global finance, supply chain and more.

Meanwhile, Vodafone has built a global data platform on GCP, including BigQuery data warehouse, data analytics, business intelligence, and machine learning, even though it is a Teradata customer, at least in Germany. Vodafone said the GCP system went live in April this year.

Snowflake, the cloud-native data warehouse startup founded in 2012, has continued to impress the market with its patented multi-cluster technology and separation of storage and compute, which the company says makes scale up and down a cinch. Snowflake has raised more than $1.4bn in venture capital.

But by taking a revenue hit as it moves to the cloud, and ensuring the market understands what it does best, there is still potential for Teradata to succeed, Olofson reckoned.

"A true enterprise data warehouse platform has certain characteristics that most of those [cloud] companies don't have. You need to be able to support things like multi-way joins, maintaining referential integrity. Some of the services they call a data warehouse are skipping some things because it makes the loads take longer or makes certain types of queries run more slowly so they just don't do them."

Even though new cloud-native data technologies can excel at discrete projects, having a view of all of an organisation's historic performance data, normalised in one place, ready for hundreds of analysts to query, is an advantage for incumbent data warehouse vendors like Teradata, he said.

"As people look at this and they say, 'Why is it that I have to extract data from all these systems and put it together into a specially designed analytic database every time I want to ask a simple question? Why can't I just ask the question?' ... After a while they realise things need to be managed. Sometimes greater agility comes from doing some advanced planning and analysis, design and construction working ahead of time.

"The core technology is what helps you to know what's going on with the business and then when something changes, what the impact is, which departments are affected and so on. You can get an immediate answer and you don't have to collect data from all over the place and load something up on a special analytic data warehouse because the data is already there. That's the whole point of a data warehouse."

By playing to its strengths, and toughing out the cloud transition, Teradata could find that the market comes back around to its way of thinking once more. It will need to make sure investors can hang on long enough, however. ®